This commit adds a new python script that can be used to print tensors
information from a tensor in a safetensors model.
The motivation for this is that during model conversion work it can
sometimes be useful to verify the shape of tensors in the original
model. While it is possible to print the tensors when loading the model
this can be slow when working with larger models.
With this script it is possible to quickly query tensor shapes.
positional arguments:
tensor_name Name of the tensor to inspect
options:
-h, --help show this help message and exit
-m MODEL_PATH, --model-path MODEL_PATH
Path to the model directory (default: MODEL_PATH environment variable)
-l, --list List unique tensor patterns in the model (layer numbers replaced with #)
```
This commit simplifies the processing of embd by removing the for loop
that currently exists which uses params.n_batch as its increment. This
commit also removes the clamping of n_eval as the size of embd is always
at most the size of params.n_batch.
The motivation is to clarify the code as it is currently a little
confusing when looking at this for loop in isolation and thinking that
it can process multiple batches.
* add an assert to verify n_eval is not greater than n_batch
Oliver Simons [Tue, 3 Feb 2026 10:33:14 +0000 (11:33 +0100)]
CUDA: Fix loop unrolling for BW in mul_mat_q_stream_k_fixup (#19053)
By providing stride_* variables as size_t (i.e., 64-bit) the compiler can
correctly unroll the [two for-loops](https://github.com/ggml-org/llama.cpp/blob/557515be1e93ed8939dd8a7c7d08765fdbe8be31/ggml/src/ggml-cuda/mmq.cuh#L3789-L3816)
on BW. This gives some perf for prefill/pp phase on BW, while not affecting
other SMs:
Gaurav Garg [Tue, 3 Feb 2026 06:41:02 +0000 (12:11 +0530)]
cuda : revert CUDA_SCALE_LAUNCH_QUEUES override until investigated (#19227)
Hangs were reported on Jetson Orin AGX if we set CUDA_SCALE_LAUNCH_QUEUES=4x. Reverting the previous PR (#19042) and updating the document to consider setting CUDA_SCALE_LAUNCH_QUEUES=4x for faster throughput on multi-GPU systems.
Sid Mohan [Mon, 2 Feb 2026 20:00:55 +0000 (12:00 -0800)]
jinja : add missing 'in' test to template engine (#19004) (#19239)
* jinja : add missing 'in' test to template engine (#19004)
The jinja template parser was missing the 'in' test from
global_builtins(), causing templates using reject("in", ...),
select("in", ...), or 'x is in(y)' to fail with
"selectattr: unknown test 'in'".
This broke tool-calling for Qwen3-Coder and any other model
whose chat template uses the 'in' test.
Added test_is_in supporting array, string, and object containment
checks, mirroring the existing 'in' operator logic in runtime.cpp.
Includes test cases for all three containment types plus
reject/select filter usage.
Co-Authored-By: Claude Opus 4.5 <redacted>
* reuse test_is_in in binary op
---------
Co-authored-by: Sid Mohan <redacted> Co-authored-by: Claude Opus 4.5 <redacted> Co-authored-by: Xuan Son Nguyen <redacted>
Matthieu Coudron [Mon, 2 Feb 2026 15:55:27 +0000 (16:55 +0100)]
server: print actual model name in 'model not found" error (#19117)
Experimenting with AI, my environment gets messy fast and it's not
always easy to know what model my software is trying to load. This helps
with troubleshooting.
before:
Error: {
code = 400,
message = "model not found",
type = "invalid_request_error"
}
After:
Error: {
code = 400,
message = "model 'toto' not found",
type = "invalid_request_error"
}
Neo Zhang [Mon, 2 Feb 2026 13:06:21 +0000 (21:06 +0800)]
Remove support for Nvidia & AMD GPU, because the oneAPI plugin for Nvidia & AMD GPU is unavailable: download/installation channels are out of work. (#19246)
User can't build up the software for Nvidia & AMD GPU.
rm the oneMath since it is only used in NV and AMD code path.
Daniel Bevenius [Mon, 2 Feb 2026 10:29:57 +0000 (11:29 +0100)]
model-conversion : add debug option to conversion script (#19265)
This commit adds a debug option to the model conversion script to enable
using the Python debugger (pdb) during model conversion.
The motivation for this is that I've found myself adding this a few
times now and it would be quicker to have this flag as an option and a
makefile target/recipe for it.
Alexis Williams [Sun, 1 Feb 2026 20:10:48 +0000 (12:10 -0800)]
nix: fix allowUnfreePredicate for packages with multiple licenses (#19237)
The allowUnfreePredicate in pkgsCuda was wrapping p.meta.license in a
list unconditionally. This fails when meta.license is already a list
of licenses, as it creates a nested list and then tries to access
.free and .shortName on the inner list.
Use lib.toList instead, which correctly handles both cases:
- Single license attrset -> wraps in list
- List of licenses -> returns unchanged
Daniele Pinna [Fri, 30 Jan 2026 20:10:24 +0000 (21:10 +0100)]
lookup, lookahead: fix crash when n_ctx not specified (#18729)
* lookup, lookahead: fix crash when n_ctx not specified
Since PR #16653 (Dec 15, 2025), the default n_ctx is 0 to enable automatic
GPU memory fitting. This causes llama-lookup and llama-lookahead to crash
when run without explicit -c flag:
Root cause: Both examples use params.n_ctx directly for batch initialization,
but params.n_ctx remains 0 even after the context is properly initialized
to n_ctx_train internally.
Bug history:
- Nov 2023: lookahead.cpp created (PR #4207) with params.n_ctx pattern
- Dec 2023: lookup.cpp created (PR #4484) with same pattern
- Nov 2024: default n_ctx changed to 4096 (PR #10136) - bug dormant
- Dec 2025: default n_ctx changed to 0 (PR #16653) - bug activated
The bug was dormant for 2+ years because params.n_ctx defaulted to 512,
then 4096. PR #16653 changed it to 0 for GPU auto-fitting, triggering
the crash.
Fix: Use llama_n_ctx(ctx) to get the actual runtime context size, matching
the pattern already used elsewhere in lookup.cpp (line 72) and in
speculative.cpp/speculative-simple.cpp.
Tested: llama-lookup now works without -c flag (12.5% acceptance on
Gemma-3-1B).
Note: llama-lookahead has a separate pre-existing issue with sequence
initialization (n_seq_max=1 vs W+G+1 needed) that is unrelated to this fix.
* lookahead: fix n_seq_max and kv_unified configuration
Lookahead decoding requires:
- W + G + 1 = 31 sequences for parallel Jacobi decoding
- Unified KV cache for coupled sequences in batch splitting
These requirements were broken after PR #14482 changed validation logic.
Consolidates fix from PR #18730 per maintainer request.
Marcello Seri [Fri, 30 Jan 2026 15:09:44 +0000 (16:09 +0100)]
jinja : add unordered_map include to value.h [no ci] (#19205)
On macos Sequoia 15.7.3, x86_64, the build has recently started failing with
```
In file included from .../code/cpp/llama.cpp/common/jinja/string.cpp:2:
.../code/cpp/llama.cpp/common/./jinja/value.h:478:10: error: no template named 'unordered_map' in namespace 'std'
478 | std::unordered_map<value, value, value_hasher, value_equivalence> unordered;
| ~~~~~^
In file included from .../code/cpp/llama.cpp/common/jinja/caps.cpp:1:
.../code/cpp/llama.cpp/common/jinja/value.h:478:10: error: no template named 'unordered_map' in namespace 'std'
478 | std::unordered_map<value, value, value_hasher, value_equivalence> unordered;
| ~~~~~^
In file included from .../code/cpp/llama.cpp/common/jinja/value.cpp:1:
In file included from .../code/cpp/llama.cpp/common/jinja/runtime.h:4:
.../code/cpp/llama.cpp/common/jinja/value.h:478:10: error: no template named 'unordered_map' in namespace 'std'
478 | std::unordered_map<value, value, value_hasher, value_equivalence> unordered;
[...]
```
After a bit of digging to make sure all the appropriate flags were used, I notifced that the necessary header was not included. This fixes the build for me and should not affect negatively other builds that for some reasons were already succeeding
Daniel Bevenius [Fri, 30 Jan 2026 09:37:06 +0000 (10:37 +0100)]
memory : remove unused tmp_buf (#19199)
This commit removes the unused tmp_buf variable from llama-kv-cache.cpp
and llama-memory-recurrent.cpp.
The tmp_buf variable was declared but never used but since it has a
non-trivial constructor/desctuctor we don't get an unused variable
warning about it.
Andrew Marshall [Thu, 29 Jan 2026 14:56:39 +0000 (09:56 -0500)]
webui: Update Svelte to fix effect_update_depth_exceeded errors (#19144)
The upstream fix is first available in 5.38.2, so constrain to at least
that version.
Rebuild pre-compiled webui index.html.gz based on these changes.
See also:
https://github.com/ggml-org/llama.cpp/issues/16347
https://github.com/huntabyte/bits-ui/issues/1687
https://github.com/sveltejs/svelte/issues/16548
This commit updates the `GraniteHybridModel` class' modify_tensors
function to properly delegate to `Mamba2Model.modify_tensors` and
`GraniteMoeModel.modify_tensors` using 'yield from' instead of 'return'.
The motivation for this is that modify_tensors is a generator function
(it uses 'yield from'), but the two calls above use return statements
but don't yield anything which means that the the caller of this
function will not receive any yielded values from it. And this causes
layer tensors to be silently dropped during conversion.
The syclcompat/math.hpp is not used anymore. The change that intrduced it was successfuly reverted (https://github.com/ggml-org/llama.cpp/pull/17826).
This include path will become obsolete and dropped in oneAPI 2026.0 effectively breaking ggml-sycl builds.
Oleksandr Kuvshynov [Wed, 28 Jan 2026 11:35:54 +0000 (06:35 -0500)]
vulkan: handle device dedup on MacOS + Vega II Duo cards (#19058)
Deduplication here relied on the fact that vulkan would return unique
UUID for different physical GPUs. It is at the moment not always the case.
On Mac Pro 2019 running Mac OS, with 2 Vega II Duo cards (so, 4 GPU total),
MotlenVK would assign same UUID to pairs of GPUs, unless they
are connected with Infinity Fabric.
See more details here: KhronosGroup/MoltenVK#2683.
The right way is to fix that in MoltenVK, but until it is fixed,
llama.cpp would only recognize 2 of 4 GPUs in such configuration.
The deduplication logic here is changed to only filter GPUs if UUID is
same but driver is different.
Daniel Bevenius [Wed, 28 Jan 2026 04:59:30 +0000 (05:59 +0100)]
sampling : remove sampling branching in output_reserve (#18811)
* sampling : remove sampling branching in output_reserve
This commit updates output_reserve in llama-context.cpp to always
allocate sampling buffers regardless of whether sampling is needed for
the current batch.
The motivation for this is to avoid reallocations and branching based on
the sampling requirements of the batch.
The XIELU operation was incorrectly using static_cast to convert
float parameters to uint32_t, which converted numeric values instead
of preserving IEEE 754 bit patterns. This caused incorrect values
to be interpreted by the GPU shader.
* Use reinterpret_cast to preserve float bit patterns when passing
through uint32_t params buffer
* Update WGSL shader parameter types from u32 to f32
* Re-enable XIELU support (was disabled due to numerical issues)
Fixes NMSE test failures for XIELU operation on WebGPU backend.
neg f16xf32xip builds and runs, havent actually ran a model that uses neg kernel yet though
Co-authored-by: James Contini <redacted> Co-authored-by: Neha Abbas <redacted> Co-authored-by: Abhijit Ramesh <redacted>
* Remove extra code and format
* Add ops documentation (finally)
* ggml webgpu: add SOFTPLUS unary operator
Implements SOFTPLUS (log(1 + exp(x))) with f16/f32 support. Uses f32
precision for intermediate calculations to prevent f16 overflow.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* Follow Vulkan backend numerical stability pattern
* ggml webgpu: add EXPM1 unary operator
Implements EXPM1 (exp(x) - 1) with f16/f32 support.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* ggml webgpu: add FLOOR unary operator
Implements FLOOR (rounds down to nearest integer) with f16/f32 support.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* ggml webgpu: add CEIL unary operator
Implements CEIL (rounds up to nearest integer) with f16/f32 support.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* ggml webgpu: add ROUND unary operator
Implements ROUND (rounds to nearest integer) with f16/f32 support.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* ggml webgpu: add TRUNC unary operator
Implements TRUNC (truncates towards zero) with f16/f32 support.
* Add shader implementation and 4 variants (f32/f16, inplace/non-inplace)
* Register pipelines and device support
* docs : update WebGPU support for unary operators (FLOOR, CEIL, ROUND, TRUNC, EXPM1, SOFTPLUS)
* Updates to webgpu get_memory
* Move shared state (webgpu_context) and device creation out of registration context, device context, and buffer context, and move into backend context
* Small cleanup
* Move Instance, Device, Adapter, Device creation, and capabilities to global state while moving Queue, pipelines, and buffers to per-thread state.
* Cleanups
* More cleanup
* Move staging_buf mutex to global context
* Resolve merge
* Resolve merge
* Resolve merge
* Clean up merge errors, delete forward declaration, and run clang-format
* Rename device_init to backend_init
* Move webgpu_context to backend_context
* Move buffer context members into global context and refactor function calls
* Run clang-format
* Remove commends
* Move parameter buffers to per-thread, add single memset_tensor param buf
* Fix CI compilation issue
* Fix builds for emscripten not supporting subgroups
Gaurav Garg [Tue, 27 Jan 2026 06:52:44 +0000 (06:52 +0000)]
[CUDA] Reduce CPU-side stalls due to the CUDA command buffer being full (#19042)
* [CUDA] Reduce CPU-side stalls due to the CUDA command buffer being full
With pipeline parallelism, during prompt processing, the CPU-side CUDA command buffer gets full, stalling the CPU. Due to this, enough work doesn't get submitted to the GPU, causing bubbles in the GPU timeline.
Fix this by setting the CUDA environment variable CUDA_SCALE_LAUNCH_QUEUES to 4x to increase the command buffer size.
* Set the env variable in the CUDA backend registry allocation
Daniel Bevenius [Tue, 27 Jan 2026 05:16:00 +0000 (06:16 +0100)]
common : clarify HTTPS build options in error message (#19103)
* common : clarify HTTPS build options in error message
This commit updates the https error message to provide clearer
instructions for users who encounter the "HTTPS is not supported" error.
The motivation for this is that it might not be clear to users that only
one of these options are needed to enable HTTPS support.
The LLAMA_OPENSSL option is also added to the message to cover all
possible build configurations.
* clarify that OpenSSL is the default for HTTPS support
overview / pkg / ggml / sources / llama.cpp / log