* server, webui: accept continue_final_message flag for vLLM API compat
Add the continue_final_message body flag from the vLLM and transformers
API. When set together with add_generation_prompt false, it triggers the
existing prefill_assistant code path, regardless of the server side
opt.prefill_assistant option. Mutual exclusion with add_generation_prompt
true is enforced, matching vLLM behavior.
WebUI sends continue_final_message and add_generation_prompt false on
the Continue button, with the matching opt in option on the chat service.
Pure API alignment, no change to the prefill logic itself. Paves the way
for the upcoming per-template prefill plumbing in common/chat.
* test: add coverage for continue_final_message vLLM compat flag
Two cases on top of the existing assistant prefill coverage. First,
continue_final_message true with add_generation_prompt false produces
the same rendered prompt as the prefill_assistant heuristic, proving
the new flag is a correct alias of the existing path. Second, both
flags set to true is rejected with HTTP 400, matching the
vLLM/transformers mutual exclusion contract.
* chore: update webui build output
* spec : refactor
* spec : drop support for incompatible vocabs
* spec : update common_speculative_init()
* cont : pass seq_id
* cont : dedup ctx_seq_rm_type
* server : sketch the ctx_dft decode loop
* server : draft prompt cache and checkpoints
* server : improve ctx names
* server, spec : transition to unified spec context
* cont : sync main and drft contexts
* cont : async drft eval when possible
* cont : handle non-ckpt models
* cont : pass correct n_past for drafting
* cont : process images throught the draft context
* spec : handle draft running out of context
* server : fix mtmd draft processing
* server : fix URL for draft model
* server : add comment
* server : clean-up + dry
* speculative-simple : update
* spec : fix n_past type
* server : fix slot ctx_drft ptr
* tools : update readme
* naming : improve consistency
* spec : refactor for multi-sequence speculative context
* cont : prepare params
* cont : prepare params
* spec : support parallel drafts
* server : support parallel drafting
* llama : reuse device buffers when possible
* server, spec : clean-up
* cont : clean-up
* cont : minor
* spec : reset `drafting` flag at the end
* spec : introduce `common_speculative_process()`
* spec : allow for multiple spec types (chain of speculators)
* replace old type field of type common_speculative_type in the
common_params_speculative struct with a vector to allow multiple
types to be specified
* introduce common_get_enabled_speculative_impls(const std::vector<enum common_speculative_type>)
to figure out which implementations the user has enabled
* introduce common_speculative_type_from_names(const std::vector<std::string> & names)
to parse the already user provided spec types
* all speculators run sequentially, best one wins (we verify its drafted tokens)
* maximize expected accepted tokens for current round by calculating the
product between the probability of accepting current token (n_acc_tokens / n_gen_drafts)
and the draft's length
---------
Co-authored-by: Petros Sideris <petros.sideris@nokia.com>
* server: support Vertex AI compatible API
* a bit safer
* support other AIP_* env var
* various fixes
* if AIP_MODE is unset, do nothing
* fix test case
* fix windows build
* chat/autoparser: the fixes
* Move optspace() to chat-peg-parser, comment out server tests invalidated due to content now allowed with forced tool calls.
* Trim whitespace on apply instead
* server: tests: fetch random media marker via /apply-template (#21962 fix)
* server: allow pinning media marker via LLAMA_MEDIA_MARKER env var
get_media_marker() checks LLAMA_MEDIA_MARKER at first call and uses it
as-is if set, falling back to the random marker otherwise.
Tests no longer need to fetch the marker dynamically via /apply-template:
the fixture sets LLAMA_MEDIA_MARKER=<__media__> so the hardcoded prompts
work as before.
Address review feedback from ngxson
* server: make get_media_marker() thread-safe via magic statics
Use a C++11 static local with a lambda initializer instead of a global
static with an empty-check. The runtime guarantees initialization exactly
once without explicit locking.
Address review feedback from ggerganov
* nits
* nits
* requirements : update transformers to 5.5.0
This commit updates the transformers dependency to version 5.5.0.
The motivation for this is that transformers 5.5.0 includes support for
Gemma4 and is required to be able to convert Gemma4 models. This is also
causing issues for user of gguf-my-repo.
Refs: https://huggingface.co/spaces/ggml-org/gguf-my-repo/discussions/202
* fix huggingface_hub version
* set version of transformers to 5.5.0
* convert : add ty ignore directives to convert_hf_to_gguf.py
This commit adds `ty: ignore` directives to transformers tokenizers
field/methods to avoid type check errors. There might be better ways to
handle this and perhaps this can be done in a follow up commit.
The motivation for this is that it looks like in transformers 5.5.0
AutoTokenizer.from_pretrained can return generic tokenizer types or None
and the type checker now produces an error when the conversion script
accesses field like tokenizer.vocab.
* convert : add ty ignore to suppress type check errors
* convert : remove incorrect type ignores
* convert : fix remaining python checks
I was running a newer version of ty locally but I've switched to
version 0.0.26 which is what CI uses and I was then able to reproduce
the errors. Sorry about the noise.
* update transformers version to 5.5.1
* fix: Bypass API Key validation for static bundle assets
* refactor: All bypassed routes in `public_endpoints`
* test: Update static assets API Key test
* common : add standard Hugging Face cache support
- Use HF API to find all files
- Migrate all manifests to hugging face cache at startup
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Check with the quant tag
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Cleanup
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Improve error handling and report API errors
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Restore common_cached_model_info and align mmproj filtering
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Prefer main when getting cached ref
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Use cached files when HF API fails
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Use final_path..
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* Check all inputs
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
---------
Signed-off-by: Adrien Gallouët <angt@huggingface.co>
* tests : fix fetch_server_test_models.py
* server: to_json_oaicompat cached_tokens
Adds OpenAI and Anthropic compatible information about the
number of cached prompt tokens used in a response.
* llama : fix pooling assertion crash in chunked GDN detection path
The chunked fused Gated Delta Net detection in sched_reserve() calls
graph_reserve(16*n_seqs, n_seqs, n_outputs, ...) where n_outputs = n_seqs.
This creates a dimension mismatch in build_pooling() for embedding models
with mean/rank pooling: build_inp_mean() creates a tensor with shape
[n_tokens=16*n_seqs, ...] while t_embd is reduced to [n_outputs=n_seqs, ...]
via out_ids, causing ggml_mul_mat to assert on ggml_can_mul_mat(a, b).
Fix: pass n_tokens as n_outputs in the chunked GDN graph reservation,
matching the pattern used by the pp/tg worst-case reservations.
Regression introduced by #20340 (d28961d).
Same class of bug as #12517, fixed by #12545.
* server : add mean pooling tests to embedding test suite
Add test_embedding_pooling_mean and test_embedding_pooling_mean_multiple
to cover the --pooling mean codepath, which was previously untested.
These tests would have caught the regression introduced by #20340 where
build_pooling() crashes with a ggml_mul_mat assertion due to mismatched
dimensions in the chunked GDN detection path.
---------
Co-authored-by: Domenico Crupi <domenico@zerovolt.it>
* Parse port numbers from MCP server URLs
* Pass scheme to http proxy for determining whether to use SSL
* Fix download on non-standard port and re-add port to logging
* add test
---------
Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
* server : support multiple model aliases via comma-separated --alias
* server : update --alias description and regenerate docs
* server : multiple model aliases and tags
- address review feedback from ngxson
- --alias accepts comma-separated values (std::set, no duplicates)
- --tags for informational metadata (not used for routing)
- aliases resolve transparently in router via get_meta/has_model
- /v1/models exposes aliases and tags fields
* regenerate docs
* nits
* server : use first alias as model_name for backward compat
address review feedback from ngxson
* server : add single-model test for aliases and tags
* from previous PR
* Make instruction(system) as first message
* Convert [input_message] (text/image/file)
* Rename convert_responses_to_chatcmpl(body) -> response_body
* Initial tool call support
* Erase instructions field from chatcmpl body
* Feed reasoning texts to chat template
* Use std::vector instead of opaque json array
* Make output_item.added events consistent
* Move `server_task_result_cmpl_partial::update` from header to source
* Match ID of output_item.added and .done events
* Add function_call only if there is no "fc_" prefix
* Add function call output at non-streaming API
* Test if ID is persistent
* Add doc
* Fix style - use trailing comma
* Rewrite state management
* catch up with upstream/master
* Fix style - "type" is the first item of SSE data
* Explicitly check "instructions" from response_body
* Make lambdas static
* Check if reasoning content exists
* Add `oai_resp_id` to task_result_state(also initialized at ctor), server_task_result_cmpl_partial, and server_task_result_cmpl_final
* Reject `input_file` since it is not supported by chatcmpl
* Add "fc_" prefix to non-straming function call id as coderabbit pointed out
---------
Co-authored-by: openingnow <>
* server : make sure children tasks are scheduled to launch with parent
* fix
* add comment pointing to this PR
* fix
* clean up
* more debug messages
* add pop_deferred_task with specific ID version
* improve the logic
* simple approach
* no double move
* correct return type of launch_slots_with_parent_task
* server : add thinking content blocks to Anthropic Messages API
Add support for returning reasoning/thinking content in Anthropic API
responses when using models with --reasoning-format deepseek and the
thinking parameter enabled.
- Non-streaming: adds thinking block before text in content array
- Streaming: emits thinking_delta events with correct block indices
- Partial streaming: tracks reasoning state across chunks via
anthropic_has_reasoning member variable
Tested with bartowski/DeepSeek-R1-Distill-Qwen-7B-GGUF model.
* server : fix Anthropic API streaming for thinking content blocks
Add signature field and fix duplicate content_block_start events in
Anthropic Messages API streaming responses for reasoning models.
* server: refactor Anthropic streaming state to avoid raw pointer
Replace raw pointer to task_result_state with direct field copies:
- Copy state fields in update() before processing chunk
- Use local copies in to_json_anthropic() instead of dereferencing
- Pre-compute state updates for next chunk in update()
This makes the data flow clearer and avoids unsafe pointer patterns.
* implement sleeping at queue level
* implement server-context suspend
* add test
* add docs
* optimization: add fast path
* make sure to free llama_init
* nits
* fix use-after-free
* allow /models to be accessed during sleeping, fix use-after-free
* don't allow accessing /models during sleep, it is not thread-safe
* fix data race on accessing props and model_meta
* small clean up
* trailing whitespace
* rm outdated comments
* backend support
* server: support multiple generations from one prompt (OAI "n" option)
* fix invalid batch
* format oai
* clean up
* disable ctx shift
* add test
* update comments
* fix style
* add n_cmpl to docs [no ci]
* allowing using both n_cmpl and n
* llama-server: add router multi-model tests (#17704)
Add 4 test cases for model router:
- test_router_unload_model: explicit model unloading
- test_router_models_max_evicts_lru: LRU eviction with --models-max
- test_router_no_models_autoload: --no-models-autoload flag behavior
- test_router_api_key_required: API key authentication
Tests use async model loading with polling and graceful skip when
insufficient models available for eviction testing.
utils.py changes:
- Add models_max, models_dir, no_models_autoload attributes to ServerProcess
- Handle JSONDecodeError for non-JSON error responses (fallback to text)
* llama-server: update test models to new HF repos
* add offline
* llama-server: fix router LRU eviction test and add preloading
Fix eviction test: load 2 models first, verify state, then load
3rd to trigger eviction. Previous logic loaded all 3 at once,
causing first model to be evicted before verification could occur.
Add module fixture to preload models via ServerPreset.load_all()
and mark test presets as offline to use cached models
* llama-server: fix split model download on Windows
---------
Co-authored-by: Xuan-Son Nguyen <thichthat@gmail.com>
* server : add Anthropic Messages API support
* remove -@pytest.mark.slow from tool calling/jinja tests
* server : remove unused code and slow/skip on test_anthropic_vision_base64_with_multimodal_model in test_anthropic_api.py
* server : removed redundant n field logic in anthropic_params_from_json
* server : use single error object instead of error_array in streaming response handler for /v1/chat/completions and use unordered_set instead of set in to_json_anthropic_stream()
* server : refactor Anthropic API to use OAI conversion
* make sure basic test always go first
* clean up
* clean up api key check, add test
---------
Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
* kv-cache : pad the size of the small SWA cache for performance
* context : pad the total context to 256
* cont : future-proof the swa pad
* server : adjust test params to new logic
* server : support unified context across slots
* cont : fix speculative decoding initialization
* context : fix n_ctx_per_seq computation
* server : purge slots one by one
* tests : add unified cache server tests
* llama : update per-seq context computation
* test-thread-safety : handle tiny training context of the input model
* server : fix server_tokens clear()
* server : use 4 slots + unified KV by default
* llama : add note about context size queries
* cont : update todos [no ci]
* context : do not cap the size of the context
* tests : adjust parameters to be CI friendlier
* context : add warning
