mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-01 09:37:35 +02:00
b10664
1140
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
6c84c7d5d8 |
model: add Qwen3.8-Flash-Next (qwen4exp) (#27742)
* gguf: add qwen4exp (Qwen3.8-Flash-Next) arch and converter
Adds the GGUF-side plumbing for HF model_type qwen4_exp:
- MODEL_ARCH.QWEN4EXP plus tensors for the low-rank hyper-connection
variant (hc_*_norm/down/up/inject) and the PLE n-gram hash embeddings.
The DeepSeek-V4 hc_*_fn/base/scale tensors are a different
parameterisation, so these are separate entries rather than reuse.
- Reuses the existing indexer, per_layer_token_embd, SSM and
compress_ratios keys unchanged.
- conversion/qwen4exp.py inherits the Qwen3.5 linear-attention V-head
reorder and interleaved mrope, concatenates the 128 PLE embedding
shards, and splits index_qk_proj into separate indexer q/k tensors.
The PLE hash multipliers reach ~2.4e13. prepare_tensors() casts every
non-float dtype to float32 before modify_tensors() runs, and GGUF array
writes infer INT32 from Python ints, so both paths are bypassed: the
constants are read from the pre-cast lazy tensors and written as
explicit UINT64 arrays.
Additive only; no existing arch changes behaviour.
* llama: load qwen4exp (Qwen3.8-Flash-Next) hparams and tensors
Adds LLM_ARCH_QWEN4EXP with its hparams and tensor loading. The graph
comes in the next commit; this makes the model load and report correct
metadata.
- hyper-connections set n_embd_out_impl = hc_count * n_embd, so the
residual stream is 4x wide and there is no output_norm: the final
mixer's hc_norm is the last norm in the model.
- registered as hybrid and given the same recurrent/attention memory
filters as Qwen3-Next and Qwen3.5.
- reuses the existing indexer, per_layer_token_embd, SSM and
compress_ratios keys as-is.
- the PLE table row count is read back from the file rather than
recomputing the vocab padding rule.
llama-model-loader gains UINT64 array support. That branch previously
threw, so no existing caller changes behaviour; it is needed because the
PLE hash multipliers do not fit in int32.
* qwen4exp: shorten comments
* llama: qwen4exp text graph with hyper-connections, GDN and MoE
Implements the decode graph for Qwen3.8-Flash-Next: the hyper-connection
residual stream, gated delta net layers, the MoE block with its gated shared
expert, and dense full attention. The QSA indexer and the PLE n-gram embedding
are not wired up yet and land in later commits.
Hyper-connections are implemented here rather than shared with deepseek4.cpp.
The two formulations agree on the [n_embd, hc, n_tokens] layout and little
else: DeepSeek-V4 mixes with a full-rank projection and Sinkhorn-normalises
it, whereas this model uses a low-rank down/silu/up sigmoid gate and collapses
by a plain mean. Only the ~10 line stream mean is genuinely common, so sharing
would mean touching DSV4's hot path and its three fused CUDA ops to reuse very
little. What is reused is the substantive part: the LLM_KV_HYPER_CONNECTION_*
keys, the n_embd_out_impl wide-residual support already in the loader, and the
layout convention.
Also allows a checkpoint to carry no PLE layers at all, which makes it
possible to bring the graph up and validate it in stages.
Validated against vLLM, the only working reference implementation. On a
scaled-down model with an init scale large enough to give non-uniform logits,
agreement with vLLM sits at the numerical noise floor: llama.cpp f32 against
its own bf16 gives 84.3% top-1 agreement over 255 positions, and this graph
against vLLM gives 85.1%. The comparison was calibrated by seeding three
deliberate bugs (silu instead of sigmoid on the delta net gate, dropping the
1/hc scale in the mix, dropping the 2x in the combine); each drops top-1 to
between 0% and 11%, an order of magnitude below the floor.
* llama: qwen4exp PLE n-gram hash embedding
Adds the per-layer embedding: a custom I32 graph input hashes each token with
its ngram_size-1 predecessors host-side and the result is a plain row gather
over the shared table, the same shape gemma3n's per-layer embedding uses. The
hash has to run on the host because the splitmix64-derived multipliers reach
2^45, so the products need 64-bit integers and an xor, neither of which ggml
has.
Predecessors that fall outside the ubatch come from a small per-sequence
history on the model, mirroring the per-request ngram_context the reference
carries. It is only trusted when contiguous with the incoming position, so a
fresh prompt or a rewound cache falls back to EOS padding rather than hashing
against stale tokens.
The depthwise conv is written out as a sum of shifted, per-channel-scaled
copies rather than through ggml_conv_1d_dw, which carries a correctness
warning upstream.
Verified two ways. The row indices match a transcription of the reference's
tensor formulation exactly, 1024 of 1024 rows, including sequences with EOS
tokens sprinkled through them to exercise the segment reset. Separately, with
PLE placed on layer 0 so its input is just the token embedding, ple_embd and
ple_gated_value match a PyTorch computation from the same checkpoint to every
printed digit.
End to end over 1023 scored positions the port sits the same distance from
vLLM with PLE as without it, 6.3 points of top-1 against 6.0, so PLE costs no
accuracy relative to the rest of the model. That common offset is vLLM's bf16
activations, which cannot be removed: its QSA kernel refuses float32.
Two bugs found along the way, both caught by the row-index check. The history
was read and updated in the same pass, so a token early in a ubatch could pick
up an earlier token of that same ubatch as prior context; it is now snapshotted
first. And an EOS token was cutting its own context, where the reference takes
the last EOS strictly before the position, so a boundary only hides tokens from
the positions after it.
Known gap: the conv carries no state across ubatches, so it is exact only for a
prefill that starts at position 0. Chunked prefill and decode need the conv
state wired into the recurrent memory, and the conv branch itself is still
numerically unverified because the fixture zeroes its weights.
* llama: carry the qwen4exp PLE conv state across ubatches
The PLE depthwise conv was zero-padding on the left, which is only right for a
prefill that starts at position 0. Decode and chunked prefill saw a truncated
history for the first (kernel-1)*ngram_size positions of every ubatch.
The PLE module sits on a layer that is also a delta-net layer, so both need a
conv history in the same recurrent row. Rather than plumb a per-layer state
size through build_rs and build_conv_state, the row is widened once and each
convolution addresses its own slice through a local helper. n_embd_r() gains
the extra span, which is zero for every other architecture because it is
derived from ple_n_heads.
Verified by feeding the same 1024 token sequence in chunks instead of one
shot: at 64 tokens per decode the logits are bit-identical to the single-shot
run, 1023 of 1023 top-1 and a maximum logprob deviation of exactly zero. At
one token per decode they differ slightly, but the no-PLE model differs more
under the same test (94.6% against 97.1%), so that is the usual gemv-versus-
gemm accumulation difference and not the state.
The conv branch is also no longer unverified. With non-zero conv weights the
port sits 6.3 points of top-1 below the numerical floor, the same distance as
with the weights zeroed and as the model with no PLE at all, so the branch
adds no error of its own.
test-llama-archs passes every existing architecture at 0.00e+00, including the
delta-net models that share this code path.
* llama: fix the qwen4exp PLE conv state and unblock test-llama-archs
build_rs writes into the state tensor in place, zeroing one row and copying the
carried-over states, so calling it twice for the same layer let the second call
clobber the first write-back. The PLE layer is also a delta-net layer, so that
is exactly what happened: both convolutions gathered the same row. They now
share a single gather per layer.
The earlier claim that the conv state was carried correctly was tested on a
fixture whose conv weights are zero, where the branch contributes nothing and
chunking matches trivially. Re-running with non-zero conv weights showed the
divergence, growing with the number of ubatch boundaries: 97.1% top-1 at one
boundary down to 90.2% at seven. With the shared gather it is bit-identical to
the single-shot run at every chunk size tried, 512, 128 and 64, with a maximum
logprob deviation of exactly zero over 1023 positions. The delta-net-only model
stays bit-identical too, so nothing regressed there.
Also derive the delta-net conv channel count the way load_arch_tensors sizes
wqkv instead of from ssm_d_inner. The two agree for this model, but n_embd_r()
only bounds the row and the convolution has to match the tensor feeding it.
test-llama-archs previously aborted on this architecture and took every later
architecture with it. qwen4exp is marked MoE-only, given the hyper-connection
keys and an ssm_d_inner consistent with its tensor derivation, and skipped for
now: the hyper-connection keys written by get_gguf_ctx are not reaching the
synthesised file, which needs a separate look. The suite completes again, 124
architectures at 0.00e+00.
* llama: optional indexer key cache in llama_memory_hybrid
Groundwork for qwen4exp's QSA sparse attention. Its indexer needs a per-token
key history for the full-attention layers, but a hybrid model cannot use
llama_kv_cache_dsa: that class derives from llama_memory_i rather than
llama_kv_cache, and llama_memory_hybrid constructs its attention cache
directly. No existing architecture pairs recurrent state with a sparse
indexer, so there was nothing to reuse wholesale.
llama_memory_hybrid therefore gains a third, optional cache, shaped the same
way llama_kv_cache_dsa shapes its lightning-indexer cache: a copy of hparams
with n_head_kv forced to 1 and n_embd_head_k_full set to indexer_head_size.
It is built only when a filter_idx callback is passed, which defaults to
nullptr, so every existing architecture gets exactly what it got before. The
per-sequence operations and the batch preparation forward to it under a null
check, matching how the DSA cache prepares its two caches over the same
ubatches.
test-llama-archs passes all 124 architectures at 0.00e+00, including the 12 in
the hybrid family that share this code. The qwen4exp fixtures are unchanged:
same logits against vLLM, and chunked evaluation still bit-identical to
single-shot.
* llama: QSA sparse attention for qwen4exp
The full-attention layers of this model do not attend to everything. An
indexer scores one mean-pooled key per block of compress_ratio tokens and
keeps a budget of the best blocks, plus the tail of tokens that do not yet
form a complete block. Below indexer_top_k + compress_ratio - 1 cached
tokens every block fits in the budget, so the result is exactly dense.
What is reused rather than rebuilt:
- the mask machinery. build_attn's DSA overload already turns a list of
token indices into a KQ mask via ggml_set_rows, so that block is lifted
out verbatim into build_attn_mask_top_k and shared with a new overload
on llm_graph_input_attn_kv. DSA's node sequence is unchanged; the new
overload exists because llama_kv_cache_dsa assumes MLA and cannot be
dropped into a hybrid model.
- the indexer key cache, which is the optional third cache added to
llama_memory_hybrid in the previous commit. It holds raw keys, because
pooling happens before the norm and the rotation.
The graph expands block scores rather than block indices: giving every
token of a block its block's score needs only a gather, where expanding
indices would need an integer multiply-add that ggml has no op for. Since
the budget is a whole number of blocks and a block's members tie exactly,
the cut still lands on a block boundary.
Everything that depends on cache layout is computed host-side in
set_input_qsa. Blocks are cuts of the position line rather than of the cell
array, so nothing assumes the cache is contiguous.
Measured on the tiny fixture against vLLM, comparing the selected token
indices directly rather than the logits:
below the budget selection identical, and 1024-token logits are
bit-identical to the pre-QSA dense path
above the budget mean jaccard 0.975
The direct index comparison is what made this correct. The reference
rectifies each head's dot product before summing over heads, which an
earlier reading of it had missed; on logits alone the resulting port looked
fine, because on a randomly initialised fixture the known-correct dense
path already disagrees with vLLM by more than the bug did. Comparing the
indices showed 0.794, and fixing the ReLU moved it to 0.975.
* llama: give the qwen4exp indexer cache the attention cache's slots
The indexer cache found its own slots, independently of the attention
cache. Both are the same size and see the same ubatches, so in a
straight-through prefill they agree, which is why every fixture and every
single-shot parity run passed. They drift once the context is being
rewritten between turns, and then the QSA top-k indices, which are applied
against the attention mask, point at the wrong cells.
The seven-turn chat test caught it on the third turn: llama-server aborted
on the assertion that the two caches report the same n_kv.
The cache is a side buffer addressed by the attention cache's cells, so it
now takes that cache's slot layout instead of computing one. Applying that
layout also marks its cells identically, so the two agree cell for cell by
construction rather than by coincidence, and the assertion can no longer
fire.
Inert where the caches already agreed: test-llama-archs green at 126 archs
and 0.00e+00, and the 4096-token tiny fixture is unchanged at max logit
delta 0.0.
* tests: record what the qwen4exp arch-test skip actually observes
The old note guessed that the hyper-connection keys never reach the file.
They do: dumping the gguf_context handed to llama_model_init_from_user
shows both among its 67 KVs, and the loader still reports one missing.
* tests: cover qwen4exp in test-llama-archs
The arch was skipped with a note guessing that the hyper-connection keys
never reached the synthesised file. They did. The suite builds a model, then
saves and reloads it, and llama_model_saver did not re-emit those keys, so
the failure was in the roundtrip leg rather than the first load. Three gaps,
all in shared code and all additive:
- add_kv_from_model wrote no hyper-connection, compress-ratio or PLE keys.
The PLE group only means anything whole, so it is written or omitted
together; the rest follow the file's existing style of writing every key
unconditionally, since an architecture that does not read one is
unaffected by a zero.
- the saver had no uint64 path at all, which the PLE hash constants need.
- add_tensors_from_model enumerates model-level tensors by hand and was
missing per_layer_tok_embd and the three final-mixer tensors.
Two smaller fixes on the qwen4exp side, both found by running the test:
- build_qsa_top_k divided by the compression ratio before asserting it was
non-zero, so a file without the key crashed instead of reporting.
- a layer with no compression ratio now falls back to dense attention,
which is what the model computes below the budget anyway. The test then
has to write a ratio to reach QSA at all, and an indexer key length no
narrower than n_rot, since the indexer ropes with the main attention's
rotary width.
Full suite: 126 archs, qwen4exp at 0.00e+00 with roundtrip OK. The tiny
fixture is unchanged, max logit delta 0.0 against the pre-QSA dense run.
* convert: stream the qwen4exp PLE table instead of concatenating it
The n-gram table arrives as 128 shards that were held in a dict and then
torch.cat-ed, so the peak was the shards plus the concatenation: around
300 GB of RSS on the real checkpoint, which rules out machines that could
otherwise convert this model.
Each shard is now written straight into a memory-mapped file at its final
row offset and dropped, so the resident set is one shard and the rest is
the page cache's problem. The temporary file sits beside the output and is
removed once the write finishes, including on failure.
Shards other than the last must be uniform for direct placement, which is
asserted rather than assumed, and a shard arriving before the stride is
known is held instead of misplaced.
Verified on the tiny fixture: the resulting GGUF is byte-identical to the
one the concatenating path produced (md5 2d274efac91ad1e9a6007efb0687e597).
* quantize: fall back to F16 for 32-block types with an odd ncols
tensor_type_fallback demotes a tensor whose ncols is not a multiple of the
target's block size, but its switch only enumerates the 256-block types. A
target that is already a 32-block type (iq4_nl, q4_0, q5_0, q8_0, ...) falls
into default: and throws, even though the function already knows how to answer
that case: the ncols check right below the switch resolves an unrepresentable
shape to F16.
Route those types into that check instead of throwing. Only paths that abort
today change, so no quantization that currently succeeds is affected.
Found on a 4-wide depthwise conv kernel. llama-quantize reported nothing but
"failed to quantize model from ...", with no tensor name and no exception text,
which made a quant recipe that had simply not pinned the tensor look like a
corrupt model. It now names the tensor and continues.
* quantize: let --tensor-type name per_layer_token_embd
per_layer_token_embd shares the TOKEN_EMBD category with token_embd.weight, so
--token-embedding-type is returned for it before any --tensor-type pattern is
consulted, and there is no way to give it a tier of its own.
That grouping is fine as a default and stays the default. It is a poor fit for
the size, though: on qwen4exp the table is 97.7 GiB of a 337.6 GiB BF16 file and
about 46% of a 4-bit one, roughly eighty times token_embd.weight, and it is
read by ggml_get_rows rather than a matmul so no imatrix ever covers it.
Allow an explicit --tensor-type pattern to name it, and only it. Nothing
changes unless such a pattern is passed, and token_embd.weight keeps the old
precedence in either case.
Measured on Qwen3.8-Flash-Next, Q4_K_M with an imatrix: the table lands at q8_0
(51.9 GiB, 113.5 GiB total) by following --token-embedding-type, and pinning it
q4_1 gives 30.5 GiB for 92.1 GiB total, 19% off the file.
* quantize: size the output buffer exactly instead of nelements * 4
The per-tensor output buffer was sized `nelements * 4`, described as an upper
bound. It is a very loose one: the output is at most 2 bytes per element
(f16/bf16) and usually well under 1.1 (q8_0 and below), so between 2x and 4x of
it is never touched. The exact size is already known here, since it is what the
quantization loop writes, what new_size sums to, and what the GGUF metadata is
asserted against a few lines later.
On a model whose largest tensor is a few GB none of this matters. On
Qwen3.8-Flash-Next it does: per_layer_token_embd is 51.2 G elements, so the
buffer was 205 GB where 54 GB is needed at q8_0 and 32 GB at q4_1.
Measured on that model, VmHWM of a live llama-quantize was 485 GB per process.
Three of them fit in 2 TB and five did not, which is what an OOM-killed quant
ladder looks like. This removes about 150 GB of that.
Byte-identical output, verified against the same binary built at the parent
commit: q4_K, q8_0, q5_K, q6_K and IQ4_XS, over BF16 and F32 sources, with and
without a PLE table present. Six cases, six matching md5s.
* qwen4exp: hash the image placeholder for multimodal batches
The PLE row indices are computed host-side from ubatch->token, and set_input
returned early when that was null. A multimodal ubatch is exactly that case:
the mtmd layer consumes the image placeholder ids and hands llama_decode
embeddings instead. The early return left the I32 index tensor uninitialised,
so ggml_get_rows indexed a 320 M row table with whatever the buffer happened to
contain, and aborted:
GGML_ASSERT(i01 >= 0 && i01 < ne01) failed
ggml_compute_forward_get_rows
mtmd_helper_decode_image_chunk -> llama_decode
Every image request crashed. Nothing caught it because the vision work had only
ever been verified by converting an mmproj, never by running one.
The reference computes the hash over input_ids, where those positions still
hold the image placeholder, so carry that id through as qwen4exp.ple.image_token_id
and hash it. The key is optional: a file converted before it existed falls back
to the PLE EOS token, which is defined and treats the image as a segment
boundary rather than crashing.
Verified end to end with llama-mtmd-cli, a Q4_K_M base and the F16 mmproj, on a
generated image with known content. The model names the red circle, the blue
square, the inverted green triangle and reads "UNSLOTH 42", each with the right
position.
* qwen4exp: support a non-unified KV cache in QSA
set_input_qsa asserted n_stream == 1, so llama-server could not serve this
model with more than one slot unless -kvu was passed. With a non-unified
cache each sequence owns its own cells, and a cell index means a different
token in each stream, so a single shared mapping is wrong.
- cell_blk, blk_cells and bias gain a stream dimension. At n_stream == 1
these collapse to the shapes they had, so the unified path is unchanged.
- Scoring is now batched over streams. ggml_mul_mat matches ne[2] on both
operands, so stream s's queries only ever meet stream s's blocks; without
this sequences would score against each other's context.
- set_input_qsa loops per stream and resolves cells through
v_cells[seq_to_stream[seq_id]], following set_input_kq_mask_impl, instead
of hardcoding v_cells[0].
- llama_kv_cache_context::get_n_stream() is added, mirroring the ns that
get_k and get_v already derive from the slot info.
build_attn_mask_top_k needed no change: it already expects
[n_top_k, n_batch, 1, n_stream], so the top-k result is reshaped to meet it.
set_input_qsa has exactly one caller, so the blast radius is qwen4exp only.
Validation, UD-Q4_K_XL on one B200:
- unified cache unchanged within noise: 1802.9/68.85 -> 1807.2/69.11 t/s at
batch 1, 2262.5/192.43 -> 2270.1/193.75 at batch 4.
- non-unified now runs at npl 1, 4, 16 where it previously aborted, and is
22% faster than the -kvu workaround at batch 16 (1205 vs 984 t/s total),
since per-stream cells avoid the cross-sequence masking a unified cache
pays for.
- no cross-stream contamination: four concurrent sequences each carrying a
distinct secret all recall their own and no other, on both cache modes.
- test-llama-archs green on qwen4exp, deepseek2, gemma3n, qwen3next, llama.
Note on testing: comparing concurrent output against solo output exactly is
not a valid check. It failed 0/4 with no bug present, and the unified-cache
control failed the same way, because batch composition changes the
floating-point reduction order and near-tied tokens flip. The contamination
test above is what the exit code gates on.
* llama: keep the qwen4exp top-k attention mask arch-local
The QSA graph needed a build_attn that attends only to the cells named by a
top_k tensor, and the first version got it by adding a llm_graph_input_attn_kv
overload to llm_graph_context and factoring the mask construction out of the
existing MLA sparse path into a shared build_attn_mask_top_k.
That put a new arch on the shared attention path and made the deepseek32 and
glm-dsa attention build depend on a helper introduced for qwen4exp. Build the
mask in src/models/qwen4exp.cpp instead and leave llama-graph.{h,cpp} exactly as
they were: the MLA path keeps its own copy of the same node sequence.
The nodes emitted are unchanged, so this is bit-identical.
* llama: hold the qwen4exp indexer cache in a new llama_memory_hybrid_idx
The indexer key cache was added by extending llama_memory_hybrid with an
optional third cache, and the host-side cell/block mapping that drives QSA was
added as set_input_qsa on llama_kv_cache. Both are shared classes that every
hybrid and every attention model goes through.
Move both into a new memory type, llama_memory_hybrid_idx, following
llama_kv_cache_msa: the indexer cache and the pos<->cell translation live with
the sparse-attention memory rather than in the classes that serve every other
architecture. llama-kv-cache.{h,cpp} and llama-memory-hybrid.{h,cpp} are
restored to their unmodified state.
init_batch is repeated from llama_memory_hybrid because the indexer cache has to
be handed the attention cache's slot infos, and those are not reachable through
the context the base returns. Allocating them separately lets the two caches
drift, which is what pointed QSA's top-k at the wrong cells before.
The context derives from llama_memory_hybrid_context so build_inp_mem_hybrid
keeps working unchanged, and get_n_stream is computed from the slot infos
exactly as llama_kv_cache_context did.
Behaviour is unchanged: logits over an 8192-token sequence are bit-identical to
the previous implementation, sparse and dense alike.
* llama: save and restore the qwen4exp indexer KV cache
llama_memory_hybrid_idx forwarded clear, seq_rm, seq_cp, seq_keep, seq_add and
seq_div to the indexer cache but not state_write / state_read, so a saved
session dropped the indexer keys and a restored one selected QSA top-k against
an empty cache. The effect is invisible until the context passes
indexer_top_k + compress_ratio - 1 cells, because QSA is exactly dense below
that and the indexer contents cannot change the result.
The indexer section is written last rather than next to the attention cache it
mirrors. As a suffix, a reader that does not expect it stops early and the
trailing bytes are caught by the size check in state_load_file; placed between
the attention and recurrent sections it would instead be parsed as recurrent
state, which can succeed and restore silent garbage. It follows the same
LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY gate as the attention cache, since a partial
checkpoint deliberately skips the token-level attention caches.
The indexer restores its own cells instead of taking the attention cache's
restored slots. The two caches share size, padding and every sequence
operation, and init_batch hands the indexer the attention cache's slot infos,
so both state_read_meta calls run find_slot over identical occupancy and land
on identical cells.
The overrides live on llama_memory_hybrid_idx, the only memory type that owns
an indexer cache, so llama_memory_hybrid and every architecture that uses it
write and read exactly the bytes they did before.
The session and sequence state versions are bumped because the qwen4exp state
layout changed. The session path already rejects a short read via its size
check, but llama_state_seq_load_file accepts one silently, so only the version
check stops a pre-fix blob from being half-restored by a fixed build.
(cherry picked from commit 2721542354f8e158c3217625f4e2e7b83e51e3fe)
* llama: make the qwen4exp PLE n-gram history per context and serialise it
The PLE hash of a token mixes in the ple_ngram_size - 1 tokens before it, which
a decode ubatch does not carry, so they were remembered in a map on
llama_model_qwen4exp. That is the wrong owner twice over.
A llama_model is shared by every context that loads it, and the map was keyed
only by llama_seq_id, so two contexts running the same sequence id - two server
instances on one model, or a draft/target pair - overwrote each other's window.
The next_pos guard turned that into EOS padding instead of a crash, so it
degraded quality silently.
The map was also in no state blob: grep found ple_hist in neither
llama-kv-cache.cpp nor llama-memory-*.cpp nor llama-context.cpp. A restored
context therefore failed the next_pos check on its first ubatch and hashed the
first tokens after the restore against EOS padding. This is why a session blob
round-tripped byte for byte while the restored context computed different
logits: the state was never in the bytes.
It moves to llama_memory_hybrid_idx, which is per context, is the memory type
qwen4exp always builds, and already does the per-sequence bookkeeping this
needs. Every sequence operation now carries the window with it:
seq_rm a rewind (p1 < 0) truncates the window to the surviving prefix and
moves next_pos to p0, so a rollback keeps exact context; a hole
punched in the middle leaves the window non-contiguous, so it is
dropped
seq_cp the destination inherits the source's window, truncated to the
copied position range - a copied sequence continues with the same
n-grams the source would have used
seq_keep every other sequence's window is dropped, like its cells
seq_add a shift that moves the whole window keeps it and moves next_pos with
it, which is the context-shift case; one that cuts through it drops
it
seq_div positions stop being consecutive, so an overlapping window is
dropped
clear everything is dropped
Dropping means next_pos = -1, which set_input turns into full EOS padding: the
same thing a fresh sequence gets, and the same thing this code did before it
followed the sequence operations at all, so no case is worse than before.
The state payload is a self-delimiting list, u32 count then per entry
{ i32 seq_id, i32 next_pos, u32 n_toks, i32 toks[n_toks] }, so a whole-context
save and a single-sequence save share one format and a single-sequence restore
can retarget the window at its destination seq_id. It is written after the
indexer section, last, for the same reason that one is: as a pure suffix an
older reader stops early instead of parsing these bytes as something else.
Unlike the indexer section it is not under LLAMA_STATE_SEQ_FLAGS_PARTIAL_ONLY.
The window is recurrent state - it is the input the PLE convolution's own
recurrent state is derived from - and the recurrent cache beside it is written
for partial checkpoints too. Gating it would leave the server's speculative
decoding checkpoints restoring the conv state without the window that produced
it.
No further version bump: LLAMA_SESSION_VERSION 10 and LLAMA_STATE_SEQ_VERSION 3
were introduced for the indexer section in the same unreleased series, and both
changes are qwen4exp-only additions to the same blob layout.
Also fixes the padding of a short window. set_input pads a window shorter than
ngram_size - 1 up to that length, but prev() indexes the snapshot with the most
recent token last, and resize() pads at the back, so the filler EOS landed where
the immediately preceding token belongs. It now pads at the front. A window is
short at a sequence start after a one-token prefill, and after a seq_rm rewind,
which the new bookkeeping makes common.
Every architecture other than qwen4exp builds llama_memory_hybrid rather than
llama_memory_hybrid_idx, has no PLE table and never asks for a history, so
nothing about its graph, its sequence operations or its state bytes changes.
(cherry picked from commit de170364c052c68fcf63285cc0028095edb9f23c)
* qwen4exp: tidy comments and simplify image token read
Rewrite the comments this series adds to the AGENTS.md rules: one or two lines,
no prose hard-wrapped mid-sentence, no narrative or history, and no comment that
only restates the code. Net 146 fewer comment lines, no code change.
Correct the PLE image comment: mtmd does not consume the placeholder ids. An
image is decoded as an embeddings-only batch, so ubatch->token is null and the
per-position ids never exist here. gemma3n and gemma4 hit the same case and
stand in row 0 of per_layer_token_embd; qwen4exp stands in the configured image
token id instead.
Read image_token_id straight from self.hparams in the converter. base.py merges
text_config into the root of hparams, and the key sits at the root of
config.json, so the config.json re-read was redundant.
(cherry picked from commit 205840c12169057da3e8d2f65ec4ceec3e18b980)
* qwen4exp: support a quantized KV cache in the QSA attention path
(cherry picked from commit 4c30574f81dc1115d08078c47b6cf8c789c0a842)
* llama: give qwen4exp a large-graph node budget
(cherry picked from commit 37c8c194e6a30e4c46ac29bee3fb264f091596ef)
* qwen4exp: drop an unused variable that breaks -Werror builds
(cherry picked from commit 528d032b51fa3cf935ed3ef6e0fb1c7401df53b5)
* quantize: dequantize and quantize large tensors in row bands
f32_conv_buf held the whole dequantized tensor, which is 204.8 GB for
per_layer_token_embd alone and dies with std::bad_alloc long before the
work buffer is reached. Dequantize and quantize in bands of whole rows
instead, capping the f32 staging at 1 GiB per band.
Rows are independent and the imatrix is indexed by column, so band
boundaries cannot change any output byte. Bands nest inside the existing
per-expert loop so each expert slice keeps its own imatrix, and a band is
kept to at least one quantization chunk per worker thread so the existing
multithreading still has work. F32 sources still stage nothing and are
banded by pointer arithmetic into the tensor.
llama_tensor_dequantize_impl now takes a first element offset; the single
caller is updated.
(cherry picked from commit 658c22549613555dbce57a772be4de8509eba3ee)
* llama: segment the qwen4exp fused QKV for tensor split
qwen4exp was missing from the gated delta net branch of get_split_segments,
so its attn_qkv.weight, shaped {n_embd, 2*key_dim + value_dim}, fell through
to the generic fused QKV rule and tripped
GGML_ASSERT(tensor->ne[axis] == n_embd + 2*n_embd_gqa) while loading with
--split-mode tensor. --split-mode layer was unaffected.
qwen4exp broadcasts K to the V heads by tiling, k_conv is grown with a plain
ggml_repeat_4d over the head axis so that v head j pairs with k head
j % n_k_heads. That is the Qwen 3.5 pattern, not the repeat interleave that
Qwen 3 Next builds explicitly, so qwen4exp takes the else branch and its V is
segmented on the scale of K.
Reported by benklop.
(cherry picked from commit 353d753f595dc81634ae6130188b31f06018f5ae)
* llama: fix the qwen4exp PLE history seq_rm(-1) iterator invalidation and the fatal-warning build
ple_hist_rm recursed over ple_hist with a range-based for and the recursive call
erases the entry it is iterating when the whole sequence is removed (p0 <= 0,
p1 < 0), so the loop then increments an invalidated iterator. It is unreachable
today only because llama_memory_recurrent::seq_rm rejects seq_id < 0 before
llama_memory_hybrid_idx::seq_rm reaches the history, which is a guard in another
class. Advance past the entry before recursing.
Two smaller things in the same area:
- the n_toks sanity bound in ple_hist_state_read was the literal 64, which is
the value of LLAMA_MAX_PLE_HEADS, not of the quantity being checked. The
window is at most ple_ngram_size - 1 tokens, so the bound is
LLAMA_MAX_PLE_NGRAM - 1, eight times tighter.
- build_conv_state_at left mem_size unused, so -DLLAMA_FATAL_WARNINGS=ON does
not compile. Predates this series; drop the line.
(cherry picked from commit 6eba44a89d5f328eb4859b844e1d28fb564cbe3e)
* qwen4exp: include llama-impl.h explicitly for llama_mul_mat_hadamard
(cherry picked from commit b634fd4d250d181ef82bf78bd00c1ae3b96a7af6)
* convert: fix the qwen4exp lint and type-check failures
flake8 flagged an unused MmprojModel import, and ty flagged seven errors in
the PLE streaming path: eos_token_id can be absent, and _ple_map, _ple_path,
_ple_row_dim and _ple_rows_per_shard are all Optional at the declaration but
were dereferenced without narrowing.
The map is opened and the stride fixed before the first shard is written, and
_finish_ple_table only runs once every shard has landed, so the invariants
hold. Assert them so the checker can see it. A missing eos_token_id now raises
with the reason instead of a TypeError from int(None).
* llama: give the qwen4exp indexer cache its own tensor names
The indexer KV cache and the attention KV cache both named their tensors
cache_k_l%d, so the Meta backend matched the indexer cache against the
attention split pattern and aborted in handle_set_rows. Tag the names
instead, and mirror the indexer cache: it has one key head and its
projections are mirrored.
(cherry picked from commit a1cdc8181134659766763a17762545a1f0e5db7b)
* qwen4exp: double the Q split granularity for tensor parallelism
qwen4exp fuses the attention gate into attn_q.weight the same way qwen3next
and qwen 3.5 do, so a device boundary must fall on a whole q+gate pair or the
Q heads stop lining up with the K/V heads and attn_output rows.
(cherry picked from commit 6c9a592f0a425a459ab6efae3b897cf68460e244)
* qwen4exp: keep the indexer cache in step across server slots
The QSA indexer keeps a side cache addressed by the cells of the attention
cache, so cell j has to hold the same token in both: the top-k indices it
produces are applied to the attention KQ mask. init_batch already hands the
indexer the attention cache's slot layout rather than letting it look for its
own, but the restore path did not. state_read called llama_kv_cache::state_read
on the two caches in turn and each ran its own find_slot over its own occupancy.
That agrees only for as long as nothing has already pushed the two caches apart,
which is the property a restore is supposed to re-establish rather than one it
can lean on.
The failure path was the worse half, and it is reachable from the public API
with nothing more than a short buffer. Truncating a good blob at 35 offsets and
feeding it to llama_state_seq_set_data left the two caches disagreeing at 5 of
them, and every one of 23 truncations of a whole-context blob did. Four of those
five land inside the attention section, so the attention cache drops the
sequence and the indexer keeps it; only the cut that lands in the indexer
section gives the opposite direction. llama_kv_cache::state_read cleans up its
own cache and rethrows, so whichever way it falls, nothing is left to bring the
two back together. The server papers over this by clearing the slot when a
prompt cache load fails; a caller of llama_state_seq_set_data that does not is
left with an indexer addressing cells that no longer mean what it thinks.
llama_kv_cache::state_read_sinfo reports the cells a restore landed in, or takes
a copy of them, and state_read_meta uses a supplied layout in place of find_slot
once it has checked that those cells are free here too. The indexer now adopts
the attention cache's restored layout by construction instead of reproducing it
by coincidence, and a layout that does not fit fails the read rather than being
applied over cells that already drifted. The hybrid restore is wrapped so that
any failure drops the sequence, or for a whole-context restore the context, from
all three caches at once, which is a state they do agree on.
* kv-cache: clear the cache once when restoring a whole context
state_read walks the streams of the cache in turn, and for a whole-context restore
each stream went through state_read_meta, which starts by calling clear(). clear()
resets every stream at once, so each stream after the first threw away the streams
already restored, and the K/V buffers with them. A non-unified cache holds one
stream per sequence, so a context saved with N sequences in it came back with only
the sequence in the last stream that carried any cells - the highest sequence id.
A unified cache has one stream and never showed it.
The cache is now emptied once, before the loop, which is what a whole-context
restore means. A blob whose streams are all empty now empties the cache as well,
where before it left the old contents in place.
* kv-cache: check the mirrored slot layout on a whole-context restore too
state_read_meta only looked at the layout it was given on the single-sequence path.
A whole-context restore lays the cells out from 0 in both caches, so they agree as
long as they restore the same number of cells, but nothing checked that they did: an
indexer section belonging to some other context was read over cells the attention
cache had filled from a different one, which is the state the indexer must never be
left in.
* qwen4exp: give the PLE conv history its own mirrored recurrent row
n_embd_r() reserved n_conv + ple_conv_state() so that one cache_r_l row could
carry both the delta-net conv state and the PLE dilated conv history, but the
QWEN4EXP arm of get_split_segments only described n_conv. Under -sm tensor the
segment sum came up short by ple_conv_state() and llama_memory_recurrent
construction aborted in ggml_backend_meta_alloc_ctx_tensors_from_buft.
Widening the segment list is not the fix. The Meta backend propagates a view's
split descriptor from its parent unchanged, so a view of one sub-range of a
split axis is sized as the whole row on every device; declaring the PLE tail as
a second segment merely moves the abort to "shape mismatch for VIEW" at graph
allocation. The two histories also want opposite policies: the delta-net state
is split by head to match wqkv and ssm_conv1d, while per_layer_tok_embd,
ple_conv1d and ple_norm_conv are all mirrored, so every device computes the
whole dilated conv and needs the whole history. One tensor cannot be both, and
the split state has no per-segment mirroring.
Move the PLE history into its own cache_ple_r_l%d row, mark it MIRRORED, and
return n_embd_r() to n_conv. The row is allocated only on layers where is_ple
holds, so mirroring one 92160-element row per device replaces a 92160-element
tail on all 36 recurrent rows: the recurrent R footprint drops rather than
grows. build_conv_state_at now takes its width from the tensor it was handed
and keys its gather on that tensor, which also drops a cont of a strided view.
* no more ple_hist (use master version)
* llama: give the qwen4exp full memory context its indexer cache
graph_reserve() walks a full memory context, and qwen4exp builds its
sparse attention only when the context exposes an indexer cache. the
full-context constructor left ctx_idx null, so the reserved worst case
was the dense fallback: a smaller graph than the one decode executes.
ggml-alloc then had to grow the compute buffer on the first decode,
past the size reported at load.
with -np 4 -c 32768 -fa on -ctk q8_0 -ctv q8_0 on an IQ1_S qwen4exp,
the reserved CUDA0 buffer was 217.00 MiB against 275.71 MiB actually
used, and CUDA_Host 42.31 MiB against 191.14 MiB. reserving the sparse
graph makes both match exactly, in unified and non-unified cache mode.
Co-authored-by: Pascal <admin@serveurperso.com>
Assisted-by: Claude
* qwen4exp: shrink the PLE hparams storage
llama_hparams is held by value inside llm_graph_params and every llm_graph_input_*,
and llm_graph_params is a stack local in graph_reserve and process_ubatch, so its
width is paid on every worker thread stack.
is_ple_impl spent 2048 bytes carrying 512 bits. It is the one per-layer flag that is
not moved through the loader's uint32 array templates, so a bitset costs nothing in
call sites and also removes the uninitialized read that non-qwen4exp archs had, since
nothing filled the array for them.
The PLE head offsets and vocab sizes are token-space indices; the gather that consumes
them already truncates to int32, so 64-bit storage was never reachable. The gguf arrays
stay uint64 for file compatibility and are narrowed on load.
sizeof(llama_hparams) 34440 -> 31944, sizeof(llm_graph_params) 34872 -> 32376.
* llama: opt-in random-access mmap advice for host-resident gather tables
qwen4exp keeps per_layer_token_embd on the host: 26.8 GiB at IQ4_NL, read
by ggml_get_rows as 16 gathers of ~90-170 bytes per token, spread across
16 head regions ~20M rows apart. Measured over 4.75M gathers, no two
consecutive gathers land on the same 4 KiB page, so the readahead the
loader asks for buys nothing here and the whole table ends up cached to
serve about 4% of itself.
llama_mmap applies POSIX_FADV_SEQUENTIAL, MAP_POPULATE and a whole-file
POSIX_MADV_WILLNEED unconditionally. Those are right for streaming the
file once into buffers and wrong for whatever stays mapped afterwards.
Under LLAMA_MMAP_RANDOM the eager pull-in is skipped and the mapping is
advised random once every tensor has been read, so the load itself keeps
its sequential readahead. That alone drops the table to 4.4% resident but
serializes one NVMe latency per gather.
The second half is what pays for it: the PLE input already computes every
row index for the ubatch before the graph runs, so the pages those rows
fall on are handed to the kernel in one batch and the reads overlap.
POSIX_MADV_WILLNEED on POSIX, PrefetchVirtualMemory on Windows, which
takes the discontiguous ranges in a single call.
Off by default and off for every other model: the batched prefetch keys
off "this mapping was advised random", which nothing sets unless the user
opts in.
-c 512 --chunks 60, cold, IQ1_S, mean of 3:
default 35.3 s 26.82 GiB resident (100%)
advice only 104.5 s 1.19 GiB resident (4.4%)
advice + prefetch 34.2 s 1.19 GiB resident (4.4%)
PPL 4.2346 +/- 0.07862 in all three. IQ1_S KLD is unchanged in every
field, including Mean KLD 0.396070 +/- 0.001931 and Same top p 77.325%.
* llama: narrow the random-access mmap advice to the gather table
The advice was applied per mapping: every mapping the model kept got
POSIX_MADV_RANDOM plus a whole-file POSIX_FADV_RANDOM, and the eager
pull-in was skipped for every file. On qwen4exp that also hit
token_embd.weight, which sits 0.33 GiB past the PLE table in the same
shard and is read densely, not by sparse gathers. Measured over
-c 512 --chunks 60 on IQ1_S it fell to 8.45% resident, against 100% with
the feature off.
A model now nominates its gather tables (qwen4exp: per_layer_tok_embd)
and only those byte ranges are advised. The range is rounded out to
whole pages, which on this model takes in 832 bytes before and 192
after. token_embd goes back to 86.55% resident and the PLE table still
drops to 4.44%; smaps shows one VM_RAND_READ VMA of exactly the table
instead of one over all 27.16 GiB that stays mapped.
posix_fadvise is dropped from the narrowed path. POSIX_FADV_RANDOM
ignores its offset and length and marks the whole open file, and the
FMODE_RANDOM it sets is only read by page_cache_sync_ra() on the read()
path, which a fault on a MADV_RANDOM vma never reaches. POSIX_FADV_
DONTNEED does take a range, so the drop mode keeps it.
The eager pull-in is now skipped only for the files holding a nominated
table, and re-issued as WILLNEED over the rest of such a file, so other
shards load exactly as before.
prefetch_rows() keys off the tensor being nominated rather than off a
mapping-level flag, so the batched readahead lands only where the advice
did.
-c 512 --chunks 60, cold, IQ1_S, mean of 3, total wall:
default 32.50 s
whole mapping 30.05 s
narrowed 30.35 s
PPL 4.2061 in all three. IQ1_S KLD is bit-identical with the feature on
and off, including Mean KLD 0.396070 +/- 0.001931 and Same top p
77.325%. tg128 73.65 +/- 0.33 narrowed against 73.49 +/- 0.34 whole.
Assisted-by: Claude
* llama: fold the random-access prefetch into its own feature flag
LLAMA_MMAP_RANDOM_PREFETCH existed to measure the two halves of the feature
apart, and the measurement is done: on a cold cache over the same wikitext
run, MADV_RANDOM without the batched readahead takes 94.4 s against 36.7 s
for an untouched mapping, while the pair together take 34.1 s. Suppressing
the kernel's readahead only pays if we replace it, so the split let a user
select a 2.6x regression through a documented switch.
Keep the accessor, since the call site reads better than a mode comparison,
but derive it from the mode alone.
* FACP (Fewer Acronym Classes Please)
* qwen4exp: bias the QSA selection per block, not per cell
The QSA bias is a graph input, so it is pinned on the host and uploaded every
decode, and at -c 32768 -np 4 its twelve copies were 768 of the 815 MiB of
reserved host compute buffer.
Only one half of it needs a cell: whether the cell sits in the always-visible
tail, and whether its block was pooled. Both are properties of the block. The
other half - empty, other sequence, or in the future - is the plain visible/not
test the attention mask already carries over the same cells, so add that mask
instead of repeating it. The bias then holds one value per block.
A block sits wholly inside or wholly outside the tail because the tail starts on
a block boundary, so one value per block is exact. Cells no block covers keep
their -inf from the mask.
The mask is F16 and the bias F32, and a mixed ggml_add reinterprets the F16
buffer as float rather than converting it, so the cast is required.
reserved host compute buffer at -c 32768 -np 4:
--kv-unified 814.86 -> 238.86 MiB, CUDA0 721.07 -> 421.07 MiB
--no-kv-unified 214.86 -> 70.86 MiB, CUDA0 317.07 -> 265.07 MiB
Selection is unchanged: over 8192 tokens, four times the budget, every QSA
layer returns identical top-k indices and the logprobs are bitwise equal.
Two things a reviewer should know. A cell whose position divides past the last
block is guarded by an assert rather than handled, because no run reached it.
And the mask's same-position M-RoPE rule cannot fire for text and was never
exercised for images, so the 2D case is unverified.
* clean up code comments
* clean up new comments
* revert LLAMA_MMAP_RANDOM
* nits
* replace some changes with #27795
* improve the m-rope image for get_prev_tokens
* LazyChunkedTensor
* fix lint
* add some validations
* reduce input nodes
* trim output tokens
* nits
* some more sanity checks
* fix llm_graph_input_ple reuse
* exclude from webgpu test
---------
Co-authored-by: danielhanchen <danielhanchen@users.noreply.github.com>
Co-authored-by: danielhanchen <unslothshared@gmail.com>
Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
Co-authored-by: Pascal <admin@serveurperso.com>
Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co>
|
||
|
|
b10f9ca58c |
spec : add DFlash2 support (local convolution + candidate selector) (#27342) (#27816)
* spec : add DFlash2 support (local convolution + candidate selector) (#27342) * support DFlash2 * Add p_min in DFlash2 Assisted-by: Claude Opus 5 * Revert unnecessary changes Assisted-by: Claude Opus 5 * Revert draft sampling in rejection sampling Assisted-by: Claude Opus 5 * Refactor code structure Assisted-by: Claude Opus 5 * Delete embedding scaling Assisted-by: Claude Opus 5 * Gate output transforms on DFlash2 Assisted-by: Claude Opus 5 * Optimize Dflash 2 cost Assisted-by: Claude Opus 5 * Avoid using atoi Assisted-by: Claude Opus 5 * Modify comments Assisted-by: Claude Opus 5 * Move llama_model_dflash_selector_top_k to llama-ext.h Assisted-by: Claude Opus 5 * Formatting Assisted-by: Claude Opus 5 * Apply patch to fix the mrope bug Assisted-by: Claude Opus 5 * fix ci Assisted-by: Claude Opus 5 * Fix graph number calculation Assisted-by: Claude Opus 5 * rename hid and unary Assisted-by: Claude Opus 5 --------- Co-authored-by: Jian Chen <jianchen0311@gmail.com> Co-authored-by: Xuan-Son Nguyen <son@huggingface.co> * revert top-k.cu changes --------- Co-authored-by: Zihan Zhang <tiancaizhangdaxian@sjtu.edu.cn> Co-authored-by: Jian Chen <jianchen0311@gmail.com> |
||
|
Xuan-Son NguyenandGitHub
|
732707dff2 | quantize: cap working memory size to avoid loading big tensors onto RAM (#27795) | ||
|
Xuan-Son NguyenandGitHub
|
fac889fb38 |
llama: model_loader: add TENSOR_READ_LAZY (#27794)
* llama: model_loader: add TENSOR_GET_ROW_LAZY * add --tensor-read-lazy * rename to TENSOR_READ_LAZY * gen docs * address comments |
||
|
|
deae5ee133 |
model : simplify MiniMax-01 graph (#27790)
Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> |
||
|
|
d7a2074112 |
models : support nanbeige4.2-3B (#27730)
Co-authored-by: admin <lizongqiang@kanzhun.com> |
||
|
Xuan-Son NguyenandGitHub
|
925e117994 |
llama: add token ID tracking to KV cell (#27762)
* kv: track token id * rm get_prev_tokens, move it to the main pr * nits * add get_prev_tokens |
||
|
Saad AliandGitHub
|
eb25b7263e |
grammar : parse \- in char classes as literal hyphen (#27591)
* grammar : accept "\-" escape in character classes gbnf_escape_char_class() escapes '-' as "\-" but parse_char() rejected that escape, so generated tool-call grammars failed to parse. Assisted-by: Claude Code <claude@anthropic.com> * tests : add parser test for "\-" in char classes Assisted-by: Claude Code <claude@anthropic.com> * tests : add integration test for "\-" in char classes Assisted-by: Claude Code <claude@anthropic.com> * tests : drop integration and parser tests |
||
|
Prabhsimran SinghandGitHub
|
a130532ae1 |
mamba2 : Flatten in/out projections to dispatch GEMM instead of GEMV (#27513)
* mamba2 : flatten mamba2 in/out projections to dispatch gemm instead of gemv * mamba2 : remove redundant output reshape |
||
|
Aman GuptaandGitHub
|
bf0a29cc16 |
Deepseek 4: -sm tensor (#26490)
* DSV4: sm tensor * set coarser granularity for head splits * fix dspark * add model saving for dsv4 + allow dflash to return on specific device * add comment about dsv4 seq_rm * simplify * add shared expert delayed allreduce * remove special test for dsv4 |
||
|
jacekpoplawskiandGitHub
|
c060ca974c | model : support MTP in GLM-4.5-Air (#26534) | ||
|
|
b0539c43ed |
DeepseekV4: fix rollback with multi-seq (#26756)
* DeepseekV4: fix rollback with multi-seq * fix model loading * make pending rollback single use * only clear cache for seq_id for full load * add assert for compress ratio * make graph topology static * pass true instead of flags in clear_compressed * cont : clean-up + TODOs --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> |
||
|
Gaurav GargandGitHub
|
d3371929bb |
[Tensor parallel] Fix meta tensor split state propagation (#27574)
* ggml : fix meta tensor split state propagation * Add test-llama-archs to CI |
||
|
Tiwei BieandGitHub
|
2115b73d8e | model : support DSpark for bailingmoe3 (#27508) | ||
|
Xuan-Son NguyenandGitHub
|
5a32f7b66e |
model: add dots3-note (#27060)
* text: conversion * init impl * address review comments * fix rope * move to a new llama_kv_cache_dsa_iswa |
||
|
Xuan-Son NguyenandGitHub
|
873e5d8e39 |
model: use ggml_rope_set_offset() (#27382)
* model: use ggml_rope_set_offset() * partially apply to deepseek2 |
||
|
Chris DanisandGitHub
|
b2e5e9b28b |
TP: enable tensor split for LFM2/LFM2MOE (#26993)
Assisted-by: deepseek-v4-flash |
||
|
Tarek DakhranandGitHub
|
07822bddf8 | model : support DSpark for LFM2 models (#27383) | ||
|
Georgi GerganovandGitHub
|
929d47a391 |
graph : create V as a view of K in the k_iswa build_attn (#27392)
build_attn with the llm_graph_input_attn_k_iswa input was using the cached K tensor itself as V. Create V as a view of K (the first v_cur->ne[0] elements of each row), like the other K-only build_attn overloads. The deepseek4 MTP call site now passes the kv tensor as v_cur. Assisted-by: pi:llama.cpp/Qwen3.8-27B |
||
|
|
7221e24f57 |
model : GraniteSWAForCausalLM / GraniteMoeSWAForCausalLM (#25505)
* feat(convert): Add conversion for GraniteSWAForCausalLM Branch: GraniteSWAForCausalLM AI-usage: full (Bob, OpenCode + Qwen3.6-35b) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat(llama): Add granite_swa support Branch: GraniteSWAForCausalLM AI-usage: full (Bob, OpenCode + Qwen3.6-35b) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat(conversion): Add conversion infra for rope_pattern array NOTE: There is other work also targeting this, so this may be removed depending on merge order. Branch: GraniteSWAForCausalLM AI-usage: full (Bob) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix(conversion): Fix SWA pattern logic and support for non-rope layers Branch: GraniteSWAForCausalLM AI-usage: full (Bob) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat(conversion): Add support for GraniteMoeSWA Branch: GraniteSWAForCausalLM AI-usage: full (Bob) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Add llama_hparams::has_rope and arch constants NOTE: This shadows the work done for Granite Speech https://github.com/ggml-org/llama.cpp/pull/25107 Branch: GraniteSWAForCausalLM AI-usage: full (Bob) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Add support for per-layer rope determination Branch: GraniteSWAForCausalLM AI-usage: full (Bob) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * style: Fix failing flake8 for extra newlines Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * test: Write out SLIDING_WINDOW_PATTERN in llama-model-saver Branch: GraniteSWAForCausalLM AI-usage: full (OpenCode + Qwen3.6-35b) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix(convert): Fix missing registration for GraniteMoeSWAForCausalLM Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Load MoE params as optional Branch: GraniteSWAForCausalLM AI-usage: draft (OpenCode + Qwen3.6-35b) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Handle MoE params in conversion branch: GraniteSWAForCausalLM AI-usage: full (OpenCode + Qwen3.6-35b) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * style: Remove unnecessary newline AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Remove unnecessary tensor additions to GRANITE architecture Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Correctly handle naming for ffn gate inp Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Always default hparams.rope_pattern to 1s This isn't strictly necessary, but it will allow other models to rely on hparams.has_rope(il) without needting to prepopulate. Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Move to has_rope for all granite model architectures Now that we have a proper hparam for this, it's better to use it and not require a hacky fallback in the hparam method itself. Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: No hacky rope_finetuned fallback in has_rope Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Fully remove rope hparam filling in granitemoe There are no granitemoe models that use NoPE (it's not actually used in the layer building below), so this was just dead code. Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Save out rope_pattern in model-saver Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Set hparams.rope_finetuned for round trip Since the value is _read_ from rope_finetuned, we need to persist it when the model is saved with the saver. Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Code review cleanup Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> * refactor: Keep gate/up fused for MoE path Branch: GraniteSWAForCausalLM AI-usage: full (Claude + Sonnet 5) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Skip GRANITE_SWA in model saver https://github.com/ggml-org/llama.cpp/pull/25505#discussion_r3773175651 Keeping is_swa_impl in the saver can break other models. Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * add sliding window pattern for model in test * style: Fix indentation Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * fix: Fix \r\n Thanks Claude! Branch: GraniteSWAForCausalLM AI-usage: none Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * feat: Keep shared expert fused Branch: GraniteSWAForCausalLM AI-usage: full (Claude + Sonnet 5) Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> * style: More indentation fixes Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> --------- Signed-off-by: Gabe Goodhart <ghart@us.ibm.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> |
||
|
Sigbjørn SkjæretandGitHub
|
2e92ecd024 | models : remove duplicate metadata load (#27378) | ||
|
Ed AddarioandGitHub
|
0596704284 |
quant : Optimise memory usage by evicting weights after processing each layer (#22877)
* Evict weights from memory after processing each layer * Revert changes * Move unmap to libllama * Unmap weights offloaded to backend * Change member's constness * Remove unmap weights offloaded to backend |
||
|
Thiago PadilhaandGitHub
|
afd439df1f |
unicode : include '~' in collapsed symbol class (#26972)
The collapsed \p{S} class was missing '~', which split " ~" into
separate pre-tokens and prevented the Ġ~ BPE merge used by DeepSeek V4.
This caused re-tokenized prompts to diverge from sampled tokens and
broke KV cache reuse.
Assisted-by: Codex
|
||
|
Sigbjørn SkjæretandGitHub
|
d8df12ebc4 | vocab : support integer tokenizer scores (#27260) | ||
|
|
9cd719af21 |
model: support speculators-format checkpoints for DSpark (#26275)
* dspark: support speculators-format checkpoints (SpecForge exports) Speculators-format DSpark drafts (e.g. SpecForge exports for the Gemma-4-26B-A4B target) differ from the dense DeepSpec checkpoints in three ways: - the config nests the backbone hparams under transformer_layer_config and gives the extract layers as aux_hidden_state_layer_ids - the block is the DFlash 1+N fill-in layout: the anchor slot is a bonus token, not a prediction slot. Written as dflash.bonus_anchor; such drafts build the block and read the mask positions exactly like DFlash (n_max drafts from a 1+n_max block), only the Markov/confidence sampling comes from DSpark - the draft output vocab may be reduced (draft_vocab_size < vocab_size) with a d2t remap table. The converter expands lm_head/markov_w2 back to the full vocab and synthesizes an lm_head bias of -1e9 on the rows the draft cannot produce, so the runtime needs no d2t remapping. Such drafts ship their own (now optional) token_embd/output tensors instead of sharing the target's Verified against gemma4-26b-a4b-dspark: greedy outputs are byte-identical with and without the draft; acceptance 0.46, mean draft len 3.7 (n_max 6). Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Fable 5 * dspark: fold the speculators draft class into DSparkModel One class now covers every DSpark variant. What used to pick the class is a single flag, because the arch name turns out to be the only thing that separates the two families: SpecForge also exports a flat schema that carries no speculators_* fields yet still uses the 1+N bonus-anchor block, so keying on those fields would silently mis-read its drafts. Also rename i0 to i_first_pred in the draft read loop and the Markov head, and give the head a real bonus_anchor bool instead of testing i0 > 0. Converting the Qwen3-8B DeepSpec draft and both gemma-4 speculators drafts produces byte-identical GGUFs. The one behaviour change is that the markov_head_type check now also covers the DeepSpec checkpoints, which previously skipped it. Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Opus 5 * dspark: address review comments - rename bonus_anchor to sample_from_anchor (GGUF key and code), matching the checkpoint config field; absent key still means anchor-first - rework the reduced draft vocab to match EAGLE3: d2t is written as I64 absolute target ids and the logits are scattered at runtime, instead of expanding lm_head/markov_w2 and synthesizing an output bias at conversion - move the t2d skip to modify_tensors, like EAGLE3 - drop _is_specforge: the arch name only picks the sample_from_anchor default, embed/lm_head sharing is decided by the draft vocab size - deduplicate the tok_embd create_tensor left behind by the rebase Verified with the RedHat gemma-4-31b speculator draft: greedy output is byte-identical with and without the draft; acceptance 0.26 (n_max 7). Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Fable 5 * dspark: fold the sample_from_anchor read into the block_size block * dspark: fix flake8 continuation indent * clean up * dspark: key the sample_from_anchor default off the export format Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Fable * dspark: drop t2d in filter_tensors Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Fable * dspark: map model.lm_head instead of bypassing the dflash prefix Co-authored-by: desovo7 <942845546@qq.com> Assisted-by: Claude Fable --------- Co-authored-by: desovo7 <942845546@qq.com> Co-authored-by: ruixiang63 <wangruixiang07@outlook.com> |
||
|
|
3733366720 |
model : BailingMoE3 Support (#26608)
* Adding support for bailingmoe3
* Adds speculative decoding support
* Make BailingMoE3 safe gate metadata optional
* bailingmoe3: apply trained SwiGLU clamps
* common: fix Bailing V3 tool argument parsing
* llama-model-saver, instantiate float vector metadata writer
* bailingmoe3: support Q-LoRA (Ling-3.0-tiny)
Ling-3.0-flash sets q_lora_rank: None and projects Q directly, so the current
implementation loads a single ATTN_Q tensor. Ling-3.0-tiny sets q_lora_rank: 256
and routes Q through a LoRA bottleneck instead:
q_a_proj -> q_a_layernorm -> q_b_proj
Conversion therefore failed with:
ValueError: Can not map tensor 'model.layers.3.attention.q_a_layernorm.weight'
Add the missing path, mirroring the existing deepseek2 MLA implementation:
* constants.py - add ATTN_Q_A / ATTN_Q_B / ATTN_Q_A_NORM to BAILINGMOE3
* tensor_mapping.py - map model.layers.{bid}.attention.q_{a,b}_proj and
q_a_layernorm
* conversion - emit attention.q_lora_rank when the config has it
* bailingmoe3.cpp - read n_lora_q; create the Q-LoRA tensors and build Q
through the bottleneck when q_lora_rank > 0
Everything is gated on q_lora_rank > 0. Ling-3.0-flash's config has no
q_lora_rank, the converter only emits the key when present, hparams.n_lora_q
defaults to 0, and get_key(..., required=false) leaves the target untouched when
the key is absent - so flash keeps taking the existing direct-Q branch.
The LoRA path produces the same shape as the direct projection, so the
nope/rope split, RoPE application and wk_b absorption downstream are unchanged.
* small mtp change
* bailingmoe3: support separate MTP GGUF and Q-LoRA MTP
* gguf: remove duplicate add_kda_gate_lower_bound definition
---------
Co-authored-by: bloomer <bloomer@booper.brushtail.me>
Co-authored-by: Dyluhn <dylanranejohnston1@gmail.com>
|
||
|
|
3cb7ffb1a1 |
model : remove some ggml_concat (#27176)
Co-authored-by: Xuan Son Nguyen <son@huggingface.co> |
||
|
|
10bf611e53 |
llama : check LoRA tensor data is within file bounds (#27056)
* llama : check LoRA tensor data is within file bounds * Update src/llama-adapter.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> |
||
|
|
ad1de39e07 |
model: add Kimi-K3 text model (#26185)
* model: add Kimi-K3 text model Hybrid KDA (linear) + MLA (full) attention as in Kimi-Linear-48B, plus five things that architecture does not have: 1. cross-layer residual attention (attn_res_block_size) 2. latent MoE (routed experts run at n_expert_latent) 3. situ activation (replaces SwiGLU everywhere) 4. MLA output gate (sigmoid gate before o_proj) 5. full-rank KDA gate (single ssm_g instead of ssm_g_a/ssm_g_b) K3's text_config reports KimiLinearForCausalLM - the older 48B architecture - so get_model_architecture routes on the top-level name instead. The KDA decay gate has two forms, selected by linear_attn_config's gate_lower_bound. It is not a clamp: when set it swaps the activation entirely (fla/ops/kda/gate.py), from -exp(A_log)*softplus(x) to lower_bound*sigmoid(exp(A_log)*x). K3 sets it to -5.0; kimi-linear leaves it unset, so that path is unchanged. Cross-layer residuals reuse ggml_dsv4_hc_pre for the weighted sum. That op is CPU + CUDA only, so Metal/Vulkan will fall back per-node until those kernels exist. The routed experts ship as compressed-tensors "mxfp4-pack-quantized". That is bit-compatible with ggml's MXFP4 - same E2M1 code assignment, same E8M0 scale byte, only the nibble positions within a block differ - so they are repacked rather than dequantized, losslessly and without a ~5.5 TB bf16 round-trip. The repack is built lazily because gguf_writer holds every added tensor until the final write. DeepSeek-V4 was already doing the identical bit-shuffling, so it now shares the helper. Verified against Moonshot's own code path (transformers + fla's Triton KDA kernels) on a tiny model exercising every K3-specific feature. Final-position logits vs the fp32 reference: 6.7e-05 rel / corr 1.00000000 for both the chunked and the recurrent delta-net path. MXFP4 blocks dequantize to the source weights with 0.0e+00 error. Assisted-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * model: fix ty errors in the Kimi-K3 converter - `_res_parts` buffers (kind, tensor) pairs, not bare tensors - `get_tensors` must return an Iterator, matching ModelBase - LazyBase's `func` takes one argument, so pass the expert loaders through `args` instead of the closure - borrowing KimiLinearModel.set_vocab from an unrelated TextModel is deliberate and safe, but not expressible in the signature No behaviour change: the MXFP4 repack still dequantizes to the source weights with 0.0e+00 error and end-to-end logits are unchanged (8.386e-03 rel, corr 0.99996630). Assisted-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * Update conversion/kimi_k3.py Co-authored-by: Boris Dvorkin <b_dvorkin@niuitmo.ru> * Increase LLAMA_MAX_EXPERTS from 512 to 1024 * tests : support for Kimi K3 in archs test * chat : add Kimi K3 chat format (reasoning, content, typed tool calls) K3's assistant output is an XTML-ish tagged format built by the template's open_tag/close_tag macros. Two properties break generic parsing: 1. The generation prompt ends with open_tag('think'), so the completion starts inside the think section with no opening marker in the output (thinking_forced_open). 2. Only <|open|>/<|close|>/<|sep|>/<|end_of_msg|> are special tokens; tag names ("think", "response", "message") are ordinary text tokens. Adds common_chat_params_init_kimi_k3 (PEG_NATIVE) with detection on the marker trio, reasoning extraction, response unwrapping, and tool-call parsing of the tools/call/argument tag structure with argument types taken from the tool schema. Includes the K3 chat template fixture and 9 test-chat cases derived from real generations of the full 2.8T model. Verified end-to-end against Kimi-K3-Q2_K (GrEarl/Kimi-K3-GGUF) on 8x B200: content, reasoning_content, streaming deltas, and tool_calls all correct; finish_reason stop/tool_calls as appropriate. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * chat : add message_delimiters for Kimi K3 Per-role message-start markers for token-level span splitting. User and assistant messages carry only the role attribute, so their full opener (through <|sep|>) is used; system and tool messages continue with more attributes (type=/tool=/index=), so those delimiters stop after the role's closing quote. Verified against the K3 tiktoken vocabulary that the closing quote is always a standalone token across all attribute variants, so the token-level prefix match stays exact. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> * fix: apply nits from @ngxson and text fixes from @danielhanchen * tests : added missing hyperparameters and tensors for Kimi K3 in test-llama-archs * chore : move overly verbose header file comments to Kimi K3 source file * tests : re-enabled KIMI_K3 in test-llama-archs for WebGPU backend * model-saver : emit kda_gate_lower_bound for Kimi K3 Quick fix. The Kimi K3 loader reads kda_gate_lower_bound and gates a graph branch on it (it scales the KDA gate when the bound is above -INFINITY), but the model saver never wrote the key, so a save->load roundtrip silently dropped it back to the -INFINITY default and changed the model's output. The real K3 config sets gate_lower_bound = -5.0. I propose to emit it from the saver, and set it to -5.0 in the test-llama-archs K3 case so the roundtrip check exercises it (the roundtrip fails without the saver line). * Refactor conditional for model architecture check * tests : re-enabled (again) KIMI_K3 and MINIMAX_M3 in test-llama-archs for WebGPU backend * fix code comments * add template on conversion * move repack_mxfp4_blocks to model base * nits * add_value_length * optimize res_stack construction * nits --------- Co-authored-by: Boris Dvorkin <b_dvorkin@niuitmo.ru> Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> Co-authored-by: Deepankar Singh <singh.deepankar39@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Caleb DeLeeuw <caleb.deleeuw@gmail.com> Co-authored-by: Xuan Son Nguyen <son@huggingface.co> |
||
|
Xuan-Son NguyenandGitHub
|
27df9199d1 |
fix: check gguf array type before reading (#27075)
* fix: check gguf array type before reading * update skill |
||
|
Fathi BoudraandGitHub
|
0177dcc730 |
common: migrate the deprecated --mmap/--no-mmap to --load-mode (#26934)
Replace the deprecated --mmap, --no-mmap, --mlock, and --direct-io flags with the unified --load-mode argument across scripts, examples, and documentation. Internal warning message and env var docs updated accordingly. Signed-off-by: Fathi Boudra <fathi.boudra@linaro.org> |
||
|
|
16d222fc5e |
model : add support for MiniMaxText01ForCausalLM and MiniMaxM1ForCausalLM (#27018)
* llama : support for MiniMax-Text-01 model * chore : renames to match the other MiniMax models * model : add logits mask as MiniMax-Text-01 embeddings tensor has zero-valued embeddings for tokens >= 200032 that produce zero logits disrupting the token sampling process * llama : replace hardcoded conditions with hparams.is_recr() * model : used build_rs() for recurrent state management * chore : code cleanup * model : optimized MiniMax-Text-01 by removing the state tranpose operations * chore : removed unnecessary ggml_cont() in MiniMax-Text-01 implementation * llama : add generic logits mask graph input * model : permuted diag_decay dimensions to avoid doing it inside MiniMax-Text-01 graph * chore : code cleanup * chore : code cleanup * model : use token positions when calculating MiniMax-Text-01 decay tensors * convert : add support for MiniMaxM1ForCausalLM as it seems to be the same as MiniMaxText01ForCausalLM * chat : add jinja template for MiniMax-M1 Co-authored-by: QscQ <qscqesze@gmail.com> * chore : code cleanup * tests : MINIMAX_01-related fixes * chore : silence Python lint errors * vocab : remove unnecessary vocab type * convert : update MiniMaxText01Model conversion to use yield when modifying tensors * convert : suppress tokens with zero-valued embeddings during MiniMax-Text-01 conversion * llama : removed logits mask - no longer necessary as token suppression is used instead * model : use common functions to make MiniMax-Text-01 implementation more concise Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> * model : use common functions to make MiniMax-Text-01 implementation more concise Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> * convert : override non-working built-in chat template during conversion * tests : skip arch MINIMAX_01 tests for WebGPU backend (it breaks again) --------- Co-authored-by: Stanisław Szymczyk <sszymczy@gmail.com> Co-authored-by: QscQ <qscqesze@gmail.com> Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> |
||
|
Xuan-Son NguyenandGitHub
|
6fed9f6ff7 |
mtmd, common: various fixes (#27071)
* apply fixes * cont * revert gguf fix |
||
|
|
1692f9e50b |
ggml : recurrent state rollback for ggml_ssm_scan (#26623)
* Initial changes for Recurrent state rollback for nemotron for cpu and cuda * Removing CPU RS rollback. Will enable it in subsequent PRs * addition of test case * Removing assert and calling runtime API to check if op is supported * removing extra API and updating the call sites for K * replace static cuda detection to runtime fused_op api * address review comments and fallback when SSM rollback not supprted * Adding changes for supporting RS-rollback in CPU. Also added test-backend-ops for cpu and cuda * removing memory manipulation as rs rollback is now supported in CPU * removing the static probe which is not needed now * correcting the format * address review comments * enabling test for all the backends, unsupported backends will fallback to CPU * Apply suggestions from code review Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> * choose different graph based on the result of fused_ssm_op is supported or not and also handled memory->n_rs_seq >1 case incase of op is not supported * Support K > 1 in ssm_scan for all backends * Fix CI Issues --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> Co-authored-by: Gaurav Garg <gaugarg@nvidia.com> |
||
|
Georgi GerganovandGitHub
|
4c1a0af40d |
llama : allow virtual igpu devices (#26953)
* llama : allow virtual igpu devices * cont : better comment |
||
|
Daniel BeveniusandGitHub
|
2bacf9ea5c |
dflash : clarify output logging of target_layer_ids (#27013)
This commit tries to make the logging of target_layer_ids a bit clearer and easier to read. Currently the output generated looks like this: ```console 0.00.468.624 D load_arch_hparams: DFlash extract_layers = [0.00.468.626 D 2, 0.00.468.626 D 6, 0.00.468.626 D 20, 0.00.468.626 D 30, 0.00.468.627 D 42, 0.00.468.627 D 520.00.468.627 D ] ``` With the changes in the commit the output will be: ```console 0.00.522.765 D load_arch_hparams: DFlash extract_layers = [2, 6, 20, 30, 42, 52] ``` |
||
|
|
680a9ae63d |
cmake : introduce semantic versioning (#26839)
* cmake : introduce semantic versioning (wip)
This commit introduces semantic versioning to llama.cpp.
* squash! cmake : introduce semantic versioning (wip)
* cmake : update test-cmake README notes [no ci]
* include libmtmd in output so show its semversioned
* ci : add make-release workflow
* ci : fix build number check in build-cmake-pkg.yml
* examples : remove trailing whitespace
* ci : abort if upstream ggml version does not exist
* ci : extract step contents into scripts
* ci : add GGML_NATIVE=OFF to ubuntu job
* examples : remove CI build information from test-cmake [no ci]
This commit removes the nightly/release information that I added
previously to keep this focused only on using building and installing
llama.cpp with cmake and being able to quickly verify changes or
troubleshoot issues.
* ci : merge scripts into single script
* remove -dev-build_number support
This commit removes the incremental build number (versioning) support
that I added. This was incorrect and we should only use the semver for
the version. Releases will be tag a nightly build and package
maintainers/managers that build from source can use the tag and it is
therefor important that the correct version is reported. So a
nightly-build will report the semver without the build number. The build
number and commit as availble via cmake and test-cmake has been updated
to include an example of using them:
```console
$ ./build.sh
[test-cmake] version: 0.1.0, build: 10360 (
|
||
|
Sigbjørn SkjæretandGitHub
|
ece98b87f7 |
model : disallow integer dflash sliding_window_pattern (#26900)
* fix sliding_window_pattern * disallow integer pattern |
||
|
ChipmunkandGitHub
|
5d9e5ac30e |
server : support slot save/restore with media inputs (#26640)
* server : save serialized image chunks at the end of the llama state * server : support multimodal slot state save/restore with packed payload * server : refine image slot state serialization * server : support media slot state and centralize media validation * server : remove unnecessary comment * server : remove defensive media checks and move the chunk type check to validate() |
||
|
|
55f453b924 |
wavtokenizer-dec : bound posnet/convnext block_count against n_layer_all (#26892)
* wavtokenizer-dec : bound posnet/convnext block_count against n_layer_all * Update src/llama-model.cpp Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> |
||
|
|
cc078b45b6 |
Dflash support for nemotron-3.5 (#26905)
* conversion: skip untrained DFlash embeddings * Add Nemotron DFlash support * Add DFlash NVFP4 support * Address review comments * add missing output_s for nvfp4 * Include change for keeping residual for last layer also if requested in future dflash models * Update conversion/qwen.py Defensive check, not needed Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> * Fixing bug introduced by merge conflict --------- Co-authored-by: Sigbjørn Skjæret <sigbjorn.skjaeret@huggingface.co> |
||
|
|
6e62ba5384 |
mtmd: support pocket-tts (#26871)
* adapt the api * text model ok * working impl, need verify and clean up * mtmd: build the pocket-tts transposed convolutions as GEMM + col2im ggml_conv_transpose_1d has no grouped mode, so the depthwise upsample was built as one convolution and one concat per channel, which floods the graph with small nodes and makes kernel launches dominate the decoder. Fold both cases into the column form the seanet decoder already needs: the general case reshapes the kernel to [IC, K * OC] and matmuls it with the input, the depthwise case batches a matmul over the channels so a step scales its own kernel. A single col2im_1d then scatter-adds the columns back to the signal, with the same shape as before, so the overlap-add tail, the streaming state and the bias are untouched. Generation time per frame drops by 80% on CUDA and by 50% on CPU. The output matches the previous implementation sample for sample, with a correlation of 0.999994 and identical frame counts. * flow_temp + frames_after_eos * chunking * mtmd: carry the remaining pocket-tts per-pack settings The language packs also tune the end-of-speech padding and the padding of short prompts, next to the temperature already carried in the mmproj: french_24l asks for 8 tail frames instead of the guessed 3, english_2026-01 asks for short prompts to be padded with spaces. Write both in the mmproj as clip.gen.audio.frames_after_eos and clip.gen.audio.pad_short_text, keyed on the pack in the conversion script like the temperature. The loader keeps them optional, so a mmproj without them behaves as before. Map semicolons to commas for every pack instead, the reference only asks for it on three of them and it costs nothing elsewhere. Existing mmproj files must be converted again to carry the two keys. On a long french text the port now lands within 2% of the reference: 22.96s against 23.44s, with the same peak level and the same amount of silence. * clip.gen.audio.model_variant * clean up code comments * nit: drop the dead flow_temp hparam, the pack table holds the default * update docs * address security problems * less invasive base.py * lint * add mtmd_gen_inp_default * add docs * rm gen_flow_temp --------- Co-authored-by: Pascal <admin@serveurperso.com> |
||
|
|
153d324bcf |
llama: add default load-mode auto, which avoids mmap on iGPUs (#26081)
* llama: add new default load-mode auto which picks mmap unless a non-Metal iGPU is used * Update ggml/src/ggml-hexagon/ggml-hexagon.cpp Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com> * set mmap_support to false on OpenCL backend * fix order of load modes * use -1 for auto * resolve load mode auto earlier to correctly pick gpu host or cpu memory * add load mode auto to llama-bench * bump virtgpu api version, regenerate docs --------- Co-authored-by: Piotr Wilkin (ilintar) <piotr.wilkin@syndatis.com> Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> |
||
|
Junmo KimandGitHub
|
14e78ddef7 |
model : fix SWA not being enabled for EXAONE 4.5 (#26848)
* model : fix SWA not being enabled for EXAONE 4.5 load_arch_hparams tests `hparams.n_layer() == 64` before LLM_KV_NEXTN_PREDICT_LAYERS has been read. n_layer() returns n_layer_all - n_layer_nextn and n_layer_nextn defaults to 0, so a GGUF carrying the MTP head (block_count=65, nextn=1) evaluates to 65 and the whole SWA block is skipped. The model type switch further down in the same function reads 64, because by then the key has been loaded. n_swa is still filled in by the unconditional get_key below the block, so llama_model_n_swa() reports 4096 and the logs look correct while only swa_type stays LLAMA_SWA_TYPE_NONE. This affects the official LGAI-EXAONE GGUF release as well. EXAONE 4.0 has no MTP head, so block_count is 64 there and the check matches. * model-loader : skip TENSOR_SKIP tensors in the metadata-only path create_tensor asserts on a null buffer type when building from metadata alone, but buft_for_tensor returns null by design for tensors marked TENSOR_SKIP, which is how architectures with nextn/MTP layers mark theirs. Those models cannot be constructed by llama_model_init_from_user at all. The file-backed path below already returns nullptr for the same tensors, so callers see the same thing either way. * tests : cover exaone4 hparams ordering Builds a synthetic exaone4 model with the layout the shipped EXAONE 4.5 GGUFs use (block_count 65 + nextn 1). The swa_type check is the one that catches the ordering bug; the n_layer_nextn and n_layer() checks only tell a broken fixture apart from a real regression. Fails before the ordering fix with "swa_type is not STANDARD", passes after. * Revert "tests : cover exaone4 hparams ordering" This reverts commit |
||
|
|
dd1ea52433 |
llama : support multi-output backend sampling (#25532)
* Enable backend sampling with token speculation * Clamp the mask sum before converting it into the sampled index * Add a numeric context parameter declaring the maximum outputs one sequence * More fixes * Don't reuse memory for output views. * Match dist between CPU and GPU * Fix CPU and backend sampling mismatches * Simpify some of the changes * Fix tests on Vulkan * More test fixes * Rebase changes * Rebase and address review comments * Address review comments * Address review comments * Update src/llama-sampler.cpp Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com> |
||
|
|
62bf73d25c |
model: Muse Glimmer Support (#26841)
* Get started with Onyx
* Add architecture
* Skip keys handled in super()
* Loading tensors
* Shorten
* Graph
* Apply suggestion from @pcuenca
* Remove norm now embedding in transformers weights
* Add eot
* Explicit output_multiplier
* Handle post_norm_eps
* No super call; unhardcode eot.
The pattern `self._set_vocab_gpt2()` seems preferred throughout the
codebase, and it allows `set_vocab()` to be called from a different part
of the Python class hierarchy: the drafter model converter that we may
need eventually.
* Register for drafting
* DFlash: inherit rope type from the linked target.
Another option would be to store it in the gguf file itself.
* mmproj conversion
Note: some fields to be renamed after the implementation works. We are
keeping compatibility with the reference Meta gguf for testing purposes.
* "clip" header declarations
* Load mmproj
* Pre-processing
* Graph
* Go back to using delimiters.
Otherwise our generations are worse.
Transformers does not use them. We need to trace inputs to verify
whether they are equivalent.
* downsample_factor -> merge_size
* Add vision graph
lol, forgot from a previous commit
* Additional renames, align with llama.cpp / transformers
* Prefer _size instead of independent _h and _w
* Fix token layout
Co-authored-by: Young Han <younghan@fb.com>
* onyx: bring the chat parser onto the onyx branch
common/chat.cpp on this branch has no Onyx handling, so a converted model
serves malformed chat: the assistant preamble leaks into content
("to=self<|message|>...") and tool calls fail with
HTTP 500 "The model produced output that does not match the expected
peg-native format"
common_chat_params_init_onyx exists on onyx-fair-patch, added there by
8bb73dd3d. It was never on this branch, so this is not a regression --
the two lines developed independently.
The code here is taken verbatim from that commit. It is the clean side of
`git merge origin/onyx-fair-patch`: chat.cpp is one of the files that
merges without conflict. The full merge is not viable -- it produces 13
conflicts, including add/add on conversion/onyx.py and src/models/onyx.cpp
where the q_norm-folding and metadata-scale approaches contradict each
other, and #4/#7 are stacked on this branch's side of that.
Verified on this branch: builds with 0 errors, converts an Onyx checkpoint,
and serving it gives "4" for "What is 2+2?" plus a correct
get_weather {"city":"Paris"} tool call, where the unported branch gives the
two failures above.
No converter or runtime changes are included, so this should not interact
with the q_norm work.
Co-authored-by: Beto de Paola <betodepaola@meta.com>
* Less params, bilinear pos-emb interpolation as a graph op instead of CPU
* Map to symbolic V_MMPROJ instead of strings
* Make a couple params explicit
* Patchify via build_inp()
* No param for rope_theta
* Small cleanup
* Restore blank line
* Unpermute, to adapt to the latest transformers checkpoint
* Apply norm after token embeddings
This follows the latest transformers approach.
* Remove duplicated function
* build_vit
* onyx: use the model rope theta on sliding-window layers
* DFlash: conversion from transformers drafter
* Revert rope_type derivation from target
NOTE: this breaks compatibility with Meta's distributed DFlash GGUFs, as
the Q/K are stored in "NEOX" (rotated half) format, like in
transformers.
* Apply suggestion from @pcuenca
* Set model type
* Remove comment that will become obsolete
* Hardcode post_norm_rms_eps instead of new param
* Derive SWA+RoPE pattern from gguf array or scalar
* Fix model type <-> number of layers
* Reorder
* Rename
* Fix typo
* DFlash: seed the draft KV cache from multimodal embedding batches
`common_speculative_impl_draft_dflash::process()` returned early on any batch carrying embeddings, so an image prefill never had its target-layer features fused through the DFlash encoder and injected into the draft's KV cache. That left a hole spanning the image's positions, and the next injection at a post-image position failed to initialize its batch:
```
decoding image batch 1/1, n_tokens_batch = 256
decode: failed to initialize batch
llama_decode: failed to decode, ret = -1
process: llama_decode(ctx_dft) failed rc=-1 (n_tokens=17, offset=0)
srv decode: failed to process speculative batch
```
Every image request with `--spec-type draft-dflash` failed with HTTP 500. Text-only was unaffected, since those batches carry token ids and were let through.
Restore the earlier condition, which admits a batch that is either tokens or embeddings and skips only the degenerate neither/both cases. The rest of `process()` is already layout-agnostic -- it gathers features via `llama_get_embeddings_layer_inp()` and indexes `batch_in.pos[]` / `batch_in.seq_id[]`, none of which assume token ids -- so this is the whole fix.
Validated against `muse-glimmer-30B-bf16.gguf` + `mmproj-muse-glimmer-30B-bf16.gguf` + a DFlash draft head, on an image describe-the-shapes request:
- before: HTTP 500, `failed to process speculative batch`
- after: HTTP 200, draft acceptance 0.34012 (167 accepted / 491 generated), mean len 3.04
Output equivalence holds, which is the property that matters: at temperature 0 the drafted response is byte-identical to the same request served with no draft attached (1213/1213 chars), so the draft is drafting correctly through the image context rather than merely not crashing.
* Conversion: prefer rewrite to mapping
* Revert "Conversion: prefer rewrite to mapping"
This reverts commit
|
||
|
|
86c298fb8a |
llama: Restore quantization of mmprojs (#26818)
* Restore quantization of mmprojs This was lost in the refactor undertaken in #22004. * add noreturn --------- Co-authored-by: Xuan Son Nguyen <son@huggingface.co> |
||
|
Ruixiang WangandGitHub
|
7a20b417f4 |
model: add MTP support for Nemotron model (#26725)
* model: add MTP support for Nemotron Nano model * model: add mtp_flags for nemotron model * address review comments |
||
|
Bar HaimandGitHub
|
157b81fe6d |
model : Granite-Switch Architecture (#25107)
* granite-switch: add llama.cpp backend (POC, CPU)
New "granite-switch" architecture: a dense, all-attention Granite-4.1
model with N embedded LoRA adapters selected per-token by control tokens.
- gguf-py schema (arch, KV keys, stacked LoRA tensor names) + writer helpers
- conversion/granite.py: GraniteSwitchModel converter (stacks N adapters +
zero base slot into per-projection A/B tensors; emits switch metadata)
- C++ arch registration (llama-arch.{h,cpp}, llama-model.{h,cpp})
- src/models/granite_switch.cpp: load + per-token switched-LoRA graph via
ggml_mul_mat_id over stacked tensors; sticky per-token index + control-token
substitution in llm_graph_input_switch::set_input
- llm_graph_input_switch in src/models/models.h
Runs end-to-end on CPU: convert 3b checkpoint (842 tensors, stacked dim 13)
and generate on both base and control-token paths. Sticky switch state is
single-sequence (POC); full multi-sequence machinery is a follow-up.
* granite-switch: add Mac (Metal) build + mid-sequence switch demo script
Self-contained script to build llama.cpp on Apple Silicon (Metal),
convert the composed 3b checkpoint, and run the crisp mid-sequence
adapter-switch demos verified on Vela:
- answerability: <|answerability|> mid-seq -> "unanswerable"
- query_rewrite: <|query_rewrite|> mid-seq -> {"rewritten_question": ...}
Each demo runs the same prompt twice, differing only by a control token
placed before the assistant turn, so the per-token switch is visible.
* granite-switch mac demo: add -no-cnv so each run is one-shot
The composed model ships a chat template, so llama-completion auto-enables
interactive conversation mode and halts at a `>` prompt after generating,
stalling the script. -no-cnv disables conversation mode: generate once from
the raw prompt and exit (also prints special tokens, making the switch visible).
* granite-switch: replace global sticky index with in-graph router attention
The POC computed the per-token adapter index on the CPU and carried it
across ubatches in ONE global `mutable int32_t poc_sticky_index`, reset
only when a ubatch contained sequence position 0. That global had two
problems:
1. Concurrency: with multiple sequences in a batch it was last-writer-
wins — one sequence's adapter leaked into the others.
2. Multi-turn: an interactive `ollama run` chat continues one KV cache,
so turn 2 never saw position 0 and the index never reset — the
adapter stayed stuck on across turns.
Port the vLLM/HF backend mechanism faithfully: a single-head causal
"router" attention recovers the adapter index in-graph. Per token, only
dim 0 carries signal — Q[0]=1, K[0]=+gain for a control token / -gain
otherwise, V[0]=adapter slot / 0 — and the causal softmax over the single
visible control token recovers that adapter's slot (readback =
clamp(round(V[0]), 0, n_adapters)). gain=15 matches config.py and is
F16-safe (no F32 cache).
The router's K/V live in the model KV cache at an extra layer
R == hparams.router_layer (== n_layer). We bump n_layer_all to n_real+1
so the cache allocator gives the router its own per-sequence slot, and
set n_layer_nextn=1 so n_layer() stays n_real — the decoder loop and
tensor loading are untouched and never reference layer R. The router K is
exempted from the k-shift RoPE loop (its dim-0 value is a literal
magnitude, not a rotation).
Because the selection now lives in the per-sequence KV cache, CONCURRENT
requests are isolated for free (problem 1 fixed; verified by
scratch/concurrent_switch_test.cpp). set_input becomes stateless pure
per-token maps; the global is gone.
Single-switch contract / known limitation, identical to vLLM & HF: the
gain is flat (no recency), so within one sequence there is no mechanism to
revert to base mid-sequence — once an adapter fires it stays on until that
sequence ends (problem 2 is therefore NOT fixed by a faithful copy; vLLM/HF
avoid it only because each served request is a fresh sequence). A client
continuing one KV cache across turns must start a fresh sequence per turn,
or opt into a recency-biased router (a deliberate divergence, not done
here). Documented in granite_switch.cpp and asserted by
scratch/multiturn_leak_test.cpp.
Verified (CPU): both demos unchanged (answerability -> "unanswerable",
query_rewrite -> rewritten query); concurrent two-sequence isolation
passes; multi-turn carry-over matches the vLLM/HF contract.
* granite-switch: drop scratch tests and mac demo for upstream PR
Remove the local-only development artifacts that should not ship in the
upstream PR:
- granite-switch-mac-demo.sh (local Metal build + demo driver)
- scratch/concurrent_switch_test.cpp
- scratch/multiturn_leak_test.cpp
Also drop the now-dangling reference to the scratch tests from the
granite_switch.cpp header comment. Leaves only the core architecture
support (conversion, gguf constants, llama-arch/model/kv-cache, and the
granite_switch graph).
* granite-switch: trim comments to match native llama.cpp style
* granite-switch: trim conversion comments to match native style
* granite-switch: drop unused adapter_ranks metadata
* granite-switch: rename arch to graniteswitch and drop obid alias
* granite-switch: fix non-ASCII comments and document router gain assumption
* granite-switch: drop section comments from constants.py to match native style
* granite-switch: add functional tensor block comments matching Granite4 Vision style
* granite-switch: clarify n_expert_used comment
State the actual constraint: mul_mat_id needs n_expert_used == 1, and
since the GGUF carries expert_count = 0 the generic loader's
n_expert == 0 => n_expert_used == 0 assertion has already passed by the
time load_arch_hparams runs, so it is forced to 1 here.
* granite-switch: note n_layer_nextn reuse has no MTP
The router carving reuses n_layer_nextn, normally the MTP/next-token
count. Clarify in the comment that it is borrowed here purely as the
trailing-layers lever and that there is no MTP head, to spare readers
the double-take.
* granite-switch: rename source file and apply review nits
* granite-switch: don't force LoRA tensors to F16, follow --outtype instead
* granite-switch: drop redundant _permute_qk wrapper, call LlamaModel.permute directly
* granite-switch: read router gain from GGUF (control_token_gain) instead of hardcoding 15.0
* granite-switch: derive n_slots()
* granite-switch: move llm_graph_input_switch into granite-switch.cpp
* granite-switch: cut AI-style narration comments
* granite-switch: collapse multi-line comments
* granite-switch: rename control_token_* maps to adapter_token_*
* granite-switch: cut noise comments
* granite-switch: rename embedded LoRA tensors to <base>.lora_a/lora_b
* granite-switch: GGML_ASSERT token input to avoid UB on embeddings
* granite-switch: TODO for raw embedding input support
* granite-switch: collapse LoRA tensor constants to .lora_a/.lora_b suffix
* granite-switch: drop n_expert_used hack, guard mul_mat_id buft probe
* granite-switch: stop forcing dense expert counts, read from config
* granite-switch: renamed control_token_gain metadata key to router_gain
* granite-switch: trim header comments to match native style
* granite-switch: collapse LoRA tensors to base name + suffix
* granite-switch: inline suffix checks in tensor op resolution
* granite-switch: drop switch-lora struct comment
* granite-switch: guard router layer index and inline n_slots
* granite-switch: group adapter metadata under {arch}.adapters.* namespace
* granite-switch: add hparams.has_rope(il) for KV-shift rope skipping
* granite-switch: skip arch in test-llama-archs (adapter fixture missing, TODO)
* granite-switch: Keys.Adapters namespace + simplify n_slots
* granite-switch: validate substitute token ids against n_vocab
* granite-switch: bound adapter count and lora rank from GGUF
* granite-switch: reject MTP context type when router_layer is set
* granite-switch: throw on bad adapter metadata instead of GGML_ASSERT
* granite-switch: use ASCII +/- in router K signal comment
* granite-switch: document n_layer_nextn repurpose and its leak points
* granite-switch: gate lora_a/lora_b op mapping on router_layer
* granite-switch: label all three preview model sizes
|
||
|
Caleb DeLeeuwandGitHub
|
0377426cef |
model-saver : fix expert shared/chunk FFN length key clobber (#26693)
The saver called add_kv with LLM_KV_EXPERT_SHARED_FEED_FORWARD_LENGTH twice, the second time passing n_ff_chexp. gguf_set_val_u32 removes-then-appends, so the second call clobbers the first: the saved shared_feed_forward_length ends up as n_ff_chexp (0 for every arch except GroveMoE), and expert_chunk_feed_forward_length is never written at all. So a save->load roundtrip of any MoE model with a shared expert loses n_ff_shexp. On reload the arch falls back to n_ff for the shexp tensor shape, that no longer matches the saved tensor, and the model FAILS to load. Hits qwen2moe, qwen3-next, granite-moe, hunyuan-moe, ernie4.5, bailingmoe2, nemotron-h, and the other shared-expert MoEs. Fix: the second call writes LLM_KV_EXPERT_CHUNK_FEED_FORWARD_LENGTH. test-llama-archs: set expert_shared_feed_forward_length to a value distinct from n_ff in the MoE setup so the roundtrip exercises it. Without the fix the reload fails on a shexp tensor-shape mismatch; with it, every arch roundtrips clean. |