]>
git.djapps.eu Git - pkg/ggml/sources/llama.cpp/log
Đinh Trọng Huy [Wed, 28 May 2025 17:01:58 +0000 (02:01 +0900)]
llama : add support for BertForSequenceClassification reranker (#13858)
* convert: add support for BertForSequenceClassification
* add support for reranking using BertForSequenceClassification
* merge checks of eos and sep
* fix lint
---------
Co-authored-by: dinhhuy <redacted>
Đinh Trọng Huy [Wed, 28 May 2025 14:34:18 +0000 (23:34 +0900)]
convert: small addition to support LlamaModel (#13838)
Co-authored-by: dinhhuy <redacted>
Sky [Wed, 28 May 2025 14:33:54 +0000 (22:33 +0800)]
server: fix remove 'image_url'/'input_audio' json-object effectlly for 'llama_params' in multimodal-model-mode (#13853)
[fix]: remove 'image_url'/'input_audio' effectlly for 'llama_params' in multimodal-model-mode
Xuan-Son Nguyen [Wed, 28 May 2025 14:12:35 +0000 (16:12 +0200)]
convert : fix qwen omni conversion (#13859)
* convert : fix qwen omni conversion
* fix typo
Alex Fanthome [Wed, 28 May 2025 13:49:28 +0000 (14:49 +0100)]
tests : change umlaut test (#11600)
Johannes Gäßler [Wed, 28 May 2025 11:33:37 +0000 (13:33 +0200)]
CUDA: fix FA tg at long context for CC >= 8.9 (#13852)
Xuan-Son Nguyen [Wed, 28 May 2025 08:05:54 +0000 (10:05 +0200)]
convert : fix tensor naming conflict for llama 4 vision (#13836)
* convert : fix tensor naming conflict for llama 4 vision
* add comment
leo-pony [Wed, 28 May 2025 03:54:20 +0000 (11:54 +0800)]
CANN: Add SOC TYPE printing in cmake configuration (#13837)
lhez [Tue, 27 May 2025 19:56:08 +0000 (12:56 -0700)]
opencl: add new ops - `argsort`, `div`, `sub`, `addrows`, `sigmoid`, `group_norm` (#13787)
* opencl: add `argsort`
* opencl: add `div`
* opencl: add `add_rows`
* opencl: add `sub`
* opencl: add `sigmoid`, both `f16` and `f32`
* opencl: add `group_norm`
lhez [Tue, 27 May 2025 19:53:14 +0000 (12:53 -0700)]
opencl: mark `mul_mat` `f32f32` as supporting non-contiguous tensors (#13790)
Jeff Bolz [Tue, 27 May 2025 16:39:07 +0000 (11:39 -0500)]
vulkan: use timestamp queries for GGML_VULKAN_PERF (#13817)
Also change it to be controlled by an env var rather than cmake flag
Georgi Gerganov [Tue, 27 May 2025 16:08:44 +0000 (19:08 +0300)]
cmake : add llama-cparams.cpp to build (#13832)
Akarshan Biswas [Tue, 27 May 2025 15:22:59 +0000 (20:52 +0530)]
SYCL: add gelu_erf kernel (#13749)
* SYCL: add gelu_erf kernel
* refactor code
Co-authored-by: Atharva Dubey <redacted>
* Use scope_op_debug_print
---------
Co-authored-by: Atharva Dubey <redacted>
Georgi Gerganov [Tue, 27 May 2025 15:04:38 +0000 (18:04 +0300)]
sync : ggml
Xuan-Son Nguyen [Tue, 27 May 2025 13:53:55 +0000 (15:53 +0200)]
ggml : add ggml_repeat_4d (#13824)
xctan [Tue, 27 May 2025 13:21:36 +0000 (21:21 +0800)]
ggml : riscv: add xtheadvector support (#13720)
* ggml : riscv: add xtheadvector support
* ggml : clean up some macro usage
Xuan-Son Nguyen [Tue, 27 May 2025 12:06:10 +0000 (14:06 +0200)]
mtmd : support Qwen 2.5 Omni (input audio+vision, no audio output) (#13784)
* mtmd : allow multiple modalities at the same time
* refactor mtmd tokenizer
* fix compile
* ok, missing SinusoidsPositionEmbedding
* first working version
* fix style
* more strict validate of n_embd
* refactor if..else to switch
* fix regression
* add test for 3B
* update docs
* fix tokenizing with add_special
* add more tests
* fix test case "huge"
* rm redundant code
* set_position_mrope_1d rm n_tokens
bandoti [Tue, 27 May 2025 11:52:40 +0000 (08:52 -0300)]
docs: remove link for llama-cli function calling (#13810)
Christian Kastner [Tue, 27 May 2025 11:18:39 +0000 (13:18 +0200)]
ggml-cpu: x86 feature detection is specific to x86 (#13811)
Diego Devesa [Tue, 27 May 2025 11:05:18 +0000 (04:05 -0700)]
ggml : allow CUDA graphs when using pipeline parallelism (#13814)
Georgi Gerganov [Tue, 27 May 2025 10:49:41 +0000 (13:49 +0300)]
kv-cells : track min/max used cells and per-sequence positions (#13808)
* kv-cells : track min/max used cells and per-sequence positions
ggml-ci
* kv-cells : fix pos-modification updates for seq_pos
ggml-ci
* kv-cells : add comments
ggml-ci
Georgi Gerganov [Tue, 27 May 2025 09:07:52 +0000 (12:07 +0300)]
sampling : make sure samplers return at least 1 token (#13822)
* sampling : min-p should always return at least one token
ggml-ci
* sampling : same for typical sampling
* tests : sampling tests use min_keep == 0
ggml-ci
Georgi Gerganov [Tue, 27 May 2025 06:40:59 +0000 (09:40 +0300)]
llama : validate seq id batch input (#13809)
* llama : validate seq id batch input
ggml-ci
* cont : fix the fix
ggml-ci
Olivier Chafik [Mon, 26 May 2025 21:34:27 +0000 (14:34 -0700)]
server: --offline mode (#13804)
* server: --offline mode (env: LLAMA_OFFLINE)
---------
Co-authored-by: Xuan-Son Nguyen <redacted>
Georgi Gerganov [Mon, 26 May 2025 19:24:01 +0000 (22:24 +0300)]
scripts : add option to compare commits in Debug (#13806)
* scripts : add option to compare commits in Debug
* cont : reuse existing CMAKE_OPTS
Georgi Gerganov [Mon, 26 May 2025 19:14:52 +0000 (22:14 +0300)]
cuda : avoid cuGetErrorString (#13791)
ggml-ci
Akarshan Biswas [Mon, 26 May 2025 15:40:36 +0000 (21:10 +0530)]
SYCL: Add non contiguous support in RMS_NORM and NORM kernels (#13611)
* SYCL: Add non contiguous input support to norm kernel
* refactor and add RMS_NORM non contiguous input support
ggml-ci
* restore subgroup reduction for multi-subgroup thread blocks in norm kernels
* Swap grid dims of nsamples and nrows
ggml-ci
* Revert "Swap grid dims of nsamples and nrows"
This reverts commit
43be2d657fec7f7fba54e2cd154106bc0fc45adf .
* restore not required changes
ggml-ci
* address review comments: change it to more like SYCL
* Use a common function to calculate offset
* remove wrap around logic for handling broadcasts
* remove static from calculate_offset fn and use ceil_div
Olivier Chafik [Mon, 26 May 2025 15:03:57 +0000 (08:03 -0700)]
server: fix streaming crashes (#13786)
* add preludes to content on partial regex match
* allow all parsers to parse non-tool-call content.
* tweak order of <|python_tag|> vs <function= parsing for functionary v3.1 format. still not ideal but hopefully less prone to crash
standby24x7 [Mon, 26 May 2025 14:55:24 +0000 (23:55 +0900)]
examples/training: Fix file name in README (#13803)
This patch fixes binary file names in README.md.
Signed-off-by: Masanari Iida <redacted>
Olivier Chafik [Mon, 26 May 2025 13:56:49 +0000 (06:56 -0700)]
`server`: fix format of streamed tool call deltas (diff name, fix id location) (#13800)
* fix deltas of tool_call.function.name
* fix tool_call.id (was in tool_call.function.id!) + add function type
* add tool_call.type
* populate empty tool_call.function.arguments on first delta
Olivier Chafik [Mon, 26 May 2025 13:16:37 +0000 (06:16 -0700)]
server: fix regression on streamed non-chat completion w/ stops (#13785)
* more forgiving message diffs: partial stop words aren't erased, full stops are
* Add (slow) server test for completion + stream + stop
Georgi Gerganov [Mon, 26 May 2025 11:03:54 +0000 (14:03 +0300)]
examples : allow extracting embeddings from decoder contexts (#13797)
ggml-ci
Georgi Gerganov [Mon, 26 May 2025 09:57:50 +0000 (12:57 +0300)]
llama : clarify deprecation message (#13794)
Romain Biessy [Mon, 26 May 2025 08:28:53 +0000 (10:28 +0200)]
sycl: Add more debug prints (#13640)
Jeff Bolz [Mon, 26 May 2025 04:02:07 +0000 (23:02 -0500)]
vulkan: mark IM2COL as supporting non-contig (#13783)
Bizhao Shi [Mon, 26 May 2025 02:20:18 +0000 (10:20 +0800)]
CANN: Add the basic supports of Flash Attention kernel (#13627)
* cann: add the basic FA support
* cann: update the readme
* cann: update the FlashAttention with PSEShift
* cann: update the input parameters in FA
* cann: update the alibi with max_bias
* cann: add the constrints of softcap
* cann: update the docs CANN.md
* cann: update the docs CANN.md
* cann: fix typo of CANN.md
* cann: add some comments and update the CANN.md
* cann: update the CANN.md
* cann: update the inner precise for fusedInferAttention
* cann: update the constraints of flash_attn_ext on ggml-cann.cpp
* cann: clean the whitespace
* cann: clean the whitespace
* cann: add a new endline
Olivier Chafik [Sun, 25 May 2025 23:30:51 +0000 (00:30 +0100)]
`server`: add `--reasoning-budget 0` to disable thinking (incl. qwen3 w/ enable_thinking:false) (#13771)
---------
Co-authored-by: ochafik <redacted>
Co-authored-by: Xuan-Son Nguyen <redacted>
Xuan-Son Nguyen [Sun, 25 May 2025 17:02:18 +0000 (19:02 +0200)]
webui : bump max upload file size to 500MB (#13779)
Sigbjørn Skjæret [Sun, 25 May 2025 14:22:29 +0000 (16:22 +0200)]
tests : improve UGM tokenizer test coverage (#13773)
Georgi Gerganov [Sun, 25 May 2025 13:34:36 +0000 (16:34 +0300)]
kv-cache : rework kv_cell (#13706)
* kv-cache : rework kv_cell
ggml-ci
* kv-cells : use "shift" instead of "delta" consistently
ggml-ci
* llama : add llama_max_parallel_sequences()
ggml-ci
* kv-cells : update comments [no ci]
* context : fail upon construction if sequences exceed max value
ggml-ci
* kv-cells : get_pos() -> pos_get() + comments
ggml-ci
* kv-cells : fix tracking of "used" cells
ggml-ci
Percy Piper [Sun, 25 May 2025 12:35:53 +0000 (13:35 +0100)]
rpc : Fix build on OpenBSD (#13541)
Xuan-Son Nguyen [Sun, 25 May 2025 12:06:32 +0000 (14:06 +0200)]
mtmd : add support for Qwen2-Audio and SeaLLM-Audio (#13760)
* mtmd : add Qwen2-Audio support
* small clean up
* update discussion link
* clarify mtmd_get_output_embd
* clarification in multimodal.md
* fix ultravox bug
* ggml_cont
ddpasa [Sun, 25 May 2025 12:04:49 +0000 (14:04 +0200)]
docs : add Moondream2 pre-quantized link (#13745)
* Multimodal: Added Moondream2 model and fixed ggml.org link
* Apply suggestions from code review
---------
Co-authored-by: name <redacted>
Co-authored-by: Xuan-Son Nguyen <redacted>
Olivier Chafik [Sun, 25 May 2025 09:45:49 +0000 (10:45 +0100)]
server: fix/test add_generation_prompt (#13770)
Co-authored-by: ochafik <redacted>
Piotr Jasiukajtis [Sun, 25 May 2025 08:29:43 +0000 (10:29 +0200)]
llama : add support for Qwen3 MoE tied word embeddings (#13768)
Akarshan Biswas [Sun, 25 May 2025 07:08:37 +0000 (12:38 +0530)]
SYCL: revert "sycl: simplify bin_bcast_kernel (#13383)" (#13752)
Temporarily reverted due to failing fp16 DIV operation
This reverts commit
02cdd2d8b092b5a4bb18e013c6887ce49ba20ac5 .
ggml-ci
Olivier Chafik [Sun, 25 May 2025 00:48:08 +0000 (01:48 +0100)]
`server`: streaming of tool calls and thoughts when `--jinja` is on (#12379)
* add common_json w/ support for truncated json healing
* add common_chat_msg_diff
* partial common_chat_parse
* refactor parser w/ optionals
* server: wire chat diffs in stream mode
* fix trigger of thinking models (must happen after thoughts are closed)
* fix functionary v3.2 raw python!
* rename: common_chat_syntax (now contains format)
* rm common_regex.at_start
* don't return empty <think></think>
* accommodate yet another deepseek r1 distill fantasy syntax (`<|tool▁calls|>`)
* fix QwQ 32B tool call parsing after thoughts (hermes2)
* better logs for grammar triggers
* consume spaces after parse_json_tool_calls
* fix required tool calls w/ thinking models that have pre-opened thinking tags
* fix thinking model's initial trigger + test qwq's template
* run most test_tool_call tests in stream + non-stream modes
* make functionary v3.2 parsing more strict (differentiate first match from others)
* send final diff from server, to close off raw python arguments
* support partial content streaming in Generic mode
* tool-call: allow content prelude before hermes2 tool calls (for Qwen2.5)
* Update function-calling.md
* Update tool_bench.py
* chat-parser: remove input from exception (llm output may contain PII)
---------
Co-authored-by: ochafik <redacted>
Co-authored-by: Olivier Chafik <redacted>
Diego Devesa [Sat, 24 May 2025 22:55:16 +0000 (15:55 -0700)]
releases : bundle llvm omp library in windows release (#13763)
Diego Devesa [Sat, 24 May 2025 20:27:03 +0000 (13:27 -0700)]
releases : enable openmp in windows cpu backend build (#13756)
Diego Devesa [Sat, 24 May 2025 20:26:47 +0000 (13:26 -0700)]
ggml-cpu : set openmp wait time if not set (#13758)
0cc4m [Sat, 24 May 2025 14:49:12 +0000 (16:49 +0200)]
Move GLM4 f32 attention fix to the correct function (#13750)
Xuan-Son Nguyen [Sat, 24 May 2025 11:06:47 +0000 (13:06 +0200)]
ggml : add ggml_gelu_erf() CUDA kernel (#13719)
* ggml : add ggml_gelu_erf() CUDA kernel
* missing semicolon
Sigbjørn Skjæret [Sat, 24 May 2025 10:29:09 +0000 (12:29 +0200)]
vocab : fix ugm tokenizer precision (#13743)
Johannes Gäßler [Sat, 24 May 2025 09:46:19 +0000 (11:46 +0200)]
CUDA: fix race condition in FA vector kernels (#13742)
Diego Devesa [Fri, 23 May 2025 20:14:00 +0000 (13:14 -0700)]
ci : enable winget package updates (#13734)
Diego Devesa [Fri, 23 May 2025 20:09:38 +0000 (13:09 -0700)]
ci : add winget package updater (#13732)
Georgi Gerganov [Fri, 23 May 2025 17:16:13 +0000 (20:16 +0300)]
hparams : initialize arrays (#13728)
ggml-ci
Xuan-Son Nguyen [Fri, 23 May 2025 15:07:04 +0000 (17:07 +0200)]
llama : allow custom list of swa_layers (#13726)
Xuan-Son Nguyen [Fri, 23 May 2025 09:03:47 +0000 (11:03 +0200)]
server : support audio input (#13714)
* server : support audio input
* add audio support on webui
Chenguang Li [Fri, 23 May 2025 08:47:53 +0000 (16:47 +0800)]
CANN: Support MUL_MAT_ID for q8_0 and q4_0 (#13705)
* [CANN]Support MUL_MAT_ID Q8 && Q4
Signed-off-by: noemotiovon <redacted>
* codestyle adjustment
Signed-off-by: noemotiovon <redacted>
---------
Signed-off-by: noemotiovon <redacted>
Xuan-Son Nguyen [Fri, 23 May 2025 06:12:48 +0000 (08:12 +0200)]
ggml : fix the order of ggml_unary_op (#13718)
Jeff Bolz [Fri, 23 May 2025 04:45:02 +0000 (00:45 -0400)]
vulkan: support CPY from any type to itself (#13695)
Reuse the f16/f32 copy shaders, and just scale the number of elements
according to the type size.
Jeff Bolz [Fri, 23 May 2025 04:33:45 +0000 (00:33 -0400)]
vulkan: Disable coopmat/coopmat2/bfloat extensions if glslc doesn't support it (#13696)
Judd [Fri, 23 May 2025 04:33:08 +0000 (12:33 +0800)]
use LOG_WARN to replace `std::cerr` (#13657)
Diego Devesa [Thu, 22 May 2025 22:21:37 +0000 (15:21 -0700)]
release : fix windows hip release (#13707)
* release : fix windows hip release
* make single hip release with multiple targets
Georgi Gerganov [Thu, 22 May 2025 19:21:07 +0000 (22:21 +0300)]
tts : fix n_ubatch + make WavTokenizer cache-less (#13713)
ggml-ci
Xuan-Son Nguyen [Thu, 22 May 2025 18:42:48 +0000 (20:42 +0200)]
mtmd : add ultravox audio input (#13623)
* convert ok, load ok
* warmup ok
* test
* still does not work?
* fix padding
* temporary give up
* fix merge conflict
* build_ultravox()
* rm test
* fix merge conflict
* add necessary mtmd APIs
* first working version (only 4s of audio)
* will this monster compile?
* fix compile
* please compile
* fPIC
* fix windows
* various fixes
* clean up audio_helpers
* fix conversion
* add some debug stuff
* long audio input ok
* adapt the api
* add --audio arg
* final touch UX
* add miniaudio to readme
* fix typo
* refactor kv metadata
* mtmd_default_marker()
Aaron Teo [Thu, 22 May 2025 18:31:29 +0000 (02:31 +0800)]
common: Include torch package for s390x (#13699)
* common: update requirements.txt to include pytorch nightly for s390x
Signed-off-by: Aaron Teo <redacted>
* common: fix torch installation via pip for s390x
Signed-off-by: Aaron Teo <redacted>
---------
Signed-off-by: Aaron Teo <redacted>
Georgi Gerganov [Thu, 22 May 2025 13:33:39 +0000 (16:33 +0300)]
server : pad small embedding batches (#13692)
ggml-ci
Sigbjørn Skjæret [Thu, 22 May 2025 12:25:05 +0000 (14:25 +0200)]
gguf-py : correct charsmap parameter typing (#13701)
Nicolò Scipione [Thu, 22 May 2025 11:54:43 +0000 (13:54 +0200)]
sycl : Remove waits from function calls (#13702)
* removes the waits in async memcpy functions
Ewan Crawford [Thu, 22 May 2025 08:24:09 +0000 (09:24 +0100)]
SYCL: Avoid using with SYCL-Graph for unsupported nodes (#13587)
Currently on a CUDA backend to SYCL when running
`GGML_SYCL_DISABLE_GRAPH=0 ./bin/test-backend-ops -b SYCL0` there
are two operations that throw an exception from the blocking
waits during queue recording.
* `-o CONCAT` : Use of blocking waits on a queue that's being recorded https://github.com/ggml-org/llama.cpp/blob/master/ggml/src/ggml-sycl/concat.cpp#L185-L187
* `-o MUL_MAT_ID`: Blocking wait on a recording queue for a copy to host memory https://github.com/ggml-org/llama.cpp/blob/master/ggml/src/ggml-sycl/ggml-sycl.cpp#L3072-L3074
We've noticed that `ggml-cuda.cu` has the
[check_node_graph_compatibility_and_refresh_copy_ops](https://github.com/ggml-org/llama.cpp/blob/
39e73ae0d69f882d7e29cecc6dd8f5052fca6731 /ggml/src/ggml-cuda/ggml-cuda.cu#L2458-L2458)
method for checking if a graph can be used, even if enabled. I've taken a
similar approach in this PR by adding a method to `ggml-sycl.cpp` for checking
if a graph can be used for the operations even if a user has asked for it to be
enabled.
Henry Linjamäki [Wed, 21 May 2025 23:21:45 +0000 (02:21 +0300)]
opencl: Add support for multiple devices (#12622)
* opencl: Add support for multiple devices
... but limited to one platform. A platform with a GPU will be preferred.
Additionally:
* Filter out devices that lack capabilities needed by the backend
implementation (half support, OpenCL 2.0+, etc).
* Make ggml_backend_opencl_reg() thread-safe.
* fixup: fix an error in sync_with_other_backends
... when there is only one OpenCL device available.
Henry Linjamäki [Wed, 21 May 2025 20:21:17 +0000 (23:21 +0300)]
opencl: fix couple crashes (#12795)
* opencl: fix couple crashes
* fix kernel launches failed on devices which do not support
non-uniform work-groups. When non-uniform work-groups are not
supported, set `local_work_size` to NULL (= let driver choose the
work-group sizes). This patch does not cover everything - just the
cases tested by test-backend-ops.
* fix sub-buffer creation failed due to `cl_buffer_region::origin` not
being aligned to `CL_DEVICE_MEM_BASE_ADDR_ALIGN`.
* OpenCL: query non-uniform WG sizes only on OpenCL 3.0+
Diego Devesa [Wed, 21 May 2025 20:09:57 +0000 (13:09 -0700)]
releases : build CPU backend separately (windows) (#13642)
Georgi Gerganov [Wed, 21 May 2025 17:00:49 +0000 (20:00 +0300)]
hparams : support models for which all layers use SWA (#13682)
ggml-ci
Georgi Gerganov [Wed, 21 May 2025 16:46:56 +0000 (19:46 +0300)]
server : improve error reporting (#13680)
antichristHater [Wed, 21 May 2025 16:40:35 +0000 (19:40 +0300)]
convert : add qwen2vl support for unsloth merges (#13686)
Sigbjørn Skjæret [Wed, 21 May 2025 14:57:38 +0000 (16:57 +0200)]
examples : switch retrieval to llama_encode (#13685)
* switch retrieval to llama_encode
* enable --no-warmup for retrieval
Emmanuel Ferdman [Wed, 21 May 2025 14:33:54 +0000 (17:33 +0300)]
gguf-py : display the invalid gguf type (#13687)
Signed-off-by: Emmanuel Ferdman <redacted>
Xuan-Son Nguyen [Wed, 21 May 2025 14:26:33 +0000 (16:26 +0200)]
ggml : add ggml_gelu_erf() (#13667)
* ggml : add ggml_gelu_na (not approximated)
* fix naming order
* rename na --> erf
* apply review suggesions
* revert naming order
Robin Davidsson [Wed, 21 May 2025 13:15:27 +0000 (15:15 +0200)]
server : Add the endpoints /api/tags and /api/chat (#13659)
* Add the endpoints /api/tags and /api/chat
Add the endpoints /api/tags and /api/chat, and improved the model metadata response
* Remove trailing whitespaces
* Removed code that is not needed for copilot to work.
Dorin-Andrei Geman [Wed, 21 May 2025 13:07:57 +0000 (16:07 +0300)]
server : fix first message identification (#13634)
* server : fix first message identification
When using the OpenAI SDK (https://github.com/openai/openai-node/blob/master/src/lib/ChatCompletionStream.ts#L623-L626) we noticed that the expected assistant role is missing in the first streaming message. Fix this by correctly checking for the first message.
Co-authored-by: Piotr Stankiewicz <redacted>
Signed-off-by: Dorin Geman <redacted>
* server : Fix checks for first role message for stream=True
Co-authored-by: Piotr Stankiewicz <redacted>
Signed-off-by: Dorin Geman <redacted>
---------
Signed-off-by: Dorin Geman <redacted>
Co-authored-by: Piotr Stankiewicz <redacted>
Georgi Gerganov [Wed, 21 May 2025 12:11:13 +0000 (15:11 +0300)]
kv-cache : simplify the interface (#13660)
* kv-cache : simplify the interface
ggml-ci
* context : revert llama_batch_allocr position change
ggml-ci
Georgi Gerganov [Wed, 21 May 2025 10:09:21 +0000 (13:09 +0300)]
model : disable SWA for Phi models (#13676)
* model : disable SWA for Phi models
ggml-ci
* model : update warning message
* model : print warning only if n_swa > 0
* model : fix typo
R0CKSTAR [Wed, 21 May 2025 01:58:49 +0000 (09:58 +0800)]
musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (#13647)
* musa: fix build warning (unused parameter)
Signed-off-by: Xiaodong Ye <redacted>
* musa: upgrade MUSA SDK version to rc4.0.1
Signed-off-by: Xiaodong Ye <redacted>
* musa: use mudnn::Unary::IDENTITY op to accelerate D2D memory copy
Signed-off-by: Xiaodong Ye <redacted>
* Update ggml/src/ggml-cuda/cpy.cu
Co-authored-by: Johannes Gäßler <redacted>
* musa: remove MUDNN_CHECK_GEN and use CUDA_CHECK_GEN instead in MUDNN_CHECK
Signed-off-by: Xiaodong Ye <redacted>
---------
Signed-off-by: Xiaodong Ye <redacted>
Co-authored-by: Johannes Gäßler <redacted>
Eve [Tue, 20 May 2025 21:35:16 +0000 (21:35 +0000)]
vulkan: fix warnings (#13626)
* small fixes
* remove ifdef
l3utterfly [Tue, 20 May 2025 16:55:30 +0000 (00:55 +0800)]
mtmd-helper : bug fix to token batching in mtmd (#13650)
* Update mtmd-helper.cpp
* Update tools/mtmd/mtmd-helper.cpp
Co-authored-by: Xuan-Son Nguyen <redacted>
---------
Co-authored-by: Xuan-Son Nguyen <redacted>
Georgi Gerganov [Tue, 20 May 2025 16:21:04 +0000 (19:21 +0300)]
model : fix llama4 graph (#13663)
ggml-ci
Georgi Gerganov [Tue, 20 May 2025 13:13:16 +0000 (16:13 +0300)]
llama : remove llama_kv_cache_view API + remove deprecated (#13653)
ggml-ci
Johannes Gäßler [Tue, 20 May 2025 12:45:07 +0000 (14:45 +0200)]
CUDA: skip fully masked-out KV in FA vec kernel (#13584)
* CUDA: skip fully masked-out KV in FA vec kernel
Sigbjørn Skjæret [Tue, 20 May 2025 10:03:17 +0000 (12:03 +0200)]
tests : avoid github urls due to throttling (#13654)
Svetlozar Georgiev [Tue, 20 May 2025 09:34:15 +0000 (10:34 +0100)]
sycl: disable reorder for sycl mulmat (#13536)
0cc4m [Tue, 20 May 2025 08:11:56 +0000 (10:11 +0200)]
Set GLM4 blk.*.attn_output.weight, kqv_out-* matmul to GGML_PREC_F32 to fix infinity values in output (#13639)
Georgi Gerganov [Tue, 20 May 2025 07:41:40 +0000 (10:41 +0300)]
metal : fix typo in FA kernel comments (#13651)
Georgi Gerganov [Tue, 20 May 2025 05:05:46 +0000 (08:05 +0300)]
kv-cache : add SWA support (#13194)
* kv-cache : prepare for SWA
ggml-ci
* kv-cache : initial iSWA implementation
ggml-ci
* kv-cache : rework error recovery logic
ggml-ci
* models : fix Phi-3 SWA parameters
ggml-ci
* model : adjust Granite to rope factor changes
ggml-ci
* server : check if context can do shifts
ggml-ci
* iswa : for now, always enable shifts (experiment)
ggml-ci
* kv-cache : simplify SWA logic
ggml-ci
* kv-cache : apply defrag when we fail to find slots for the batch
ggml-ci
* llama : update docs about llama_decode
ggml-ci
* kv-cache : update warning logs when no space for the batch is available
ggml-ci
* llama : add llama_kv_self_seq_pos_min()
* kv-cache : keep track of partial SWA computes and print warnings
* server : disallow use cases involving partial SWA context
ggml-ci
* llama : add param to control SWA cache size
ggml-ci
* minor : clean-up
ggml-ci
Xinpeng Dou [Tue, 20 May 2025 03:43:43 +0000 (11:43 +0800)]
CANN: Update CANN model support (#13162)
* Update CANN model support status
* Update of model support
* update
* update
* update
* fix format of CANN.md
* fix format of CANN.md
* fix format of CANN.md
Nicolò Scipione [Tue, 20 May 2025 00:54:43 +0000 (02:54 +0200)]
sycl : Overcoming workaround for mmap() allocation on Windows (#13482)
* Remove mmap workaround on windows
After some testing I found that mmap is supported on windows and for
many GPUs on Linux. Therefore I remove the workaround for windows since
it is not necessary.
* Update llama-bench README
SYCL backend introduced a workaround that allows execution of
llama-bench also without specifying `--mmp 0` flag
psocolovsky [Mon, 19 May 2025 19:17:36 +0000 (21:17 +0200)]
common : add load_progress_callback (#13617)
0cc4m [Mon, 19 May 2025 15:54:08 +0000 (17:54 +0200)]
Vulkan: Add f32 accumulator support to quantized mul mat to fix GLM4 32B incoherence (#13607)