Commits · ef3027bd7c460f16d67a85d92150833a296a59c8 · OpenDAS / dynamo

"lib/llm/vscode:/vscode.git/clone" did not exist on "a9c0e0c7fa7adfaa8276227dbd020a4e919da46b"

12 Dec, 2025 1 commit

feat: DeepSeek V3.2 chat template support (#4797) · 1efc7d63

Vladislav Nosivskoy authored Dec 12, 2025


Signed-off-by: Vladislav Nosivskoy <vladnosiv@gmail.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

1efc7d63

11 Dec, 2025 1 commit

feat: add cached tokens prometheus metric (#4534) · f8bb53c0

Vladislav Nosivskoy authored Dec 11, 2025


Signed-off-by: Vladislav Nosivskoy <vladnosiv@gmail.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

f8bb53c0

09 Dec, 2025 1 commit
- feat: add tool_choice support (#4722) · 5585f803
  Vladislav Nosivskoy authored Dec 09, 2025
```
Signed-off-by: Vladislav Nosivskoy <vladnosiv@gmail.com>
```
  5585f803
04 Dec, 2025 1 commit
- fix: parallel tool call indices in streaming (#4723) · 8601ccdb
  Vladislav Nosivskoy authored Dec 04, 2025
```
Signed-off-by: Vladislav Nosivskoy <vladnosiv@gmail.com>
```
  8601ccdb
03 Dec, 2025 1 commit

feat: add Kserve readiness endpoint support: ServerLive / ServerReady / ModelReady (#4708) · d90dc39d

GuanLuo authored Dec 03, 2025


Signed-off-by: Guan Luo <gluo@nvidia.com>
Signed-off-by: GuanLuo <41310872+GuanLuo@users.noreply.github.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

d90dc39d

02 Dec, 2025 3 commits

fix: ModelDeploymentCard obtains full set of eos_token_ids by taking union... · 4ace4c85

GuanLuo authored Dec 02, 2025

fix: ModelDeploymentCard obtains full set of eos_token_ids by taking union from different files (#3192)
Signed-off-by: Guan Luo <gluo@nvidia.com>
Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>

4ace4c85

refactor: New config types for tool calls (#4575) · 44e8600a

William Zhang authored Dec 02, 2025

* Why?

We would like the ability to configure different parser types. Prior to
this commit, only the JSON parser could be configured.

* What?

This commit refactors the tool parser config in the following ways:
- the `format` and `json` fields of `ToolParserConfig` are merged into
  a single `config` field that is a "discriminated union" type. Each
  parser type can declare its own configuration options.
- a `XmlParserConfig` is defined with a default factory method that
  corresponds to the Qwen3 coder configuration.
- affected calls and tests are adjusted.

44e8600a

feat: lora - centralize lora cache key, restructure folders, s3 resiliency (#4644) · 71f94eda
Biswa Panda authored Dec 02, 2025

71f94eda

25 Nov, 2025 1 commit
- refactor(llm): Rename EngineConfig::Static to InProcess (#4585) · 0fc5273c
  Graham King authored Nov 25, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  0fc5273c
21 Nov, 2025 2 commits
- chore: Make nats_client private at crate level, various tidy up (#4513) · f05f7629
  Graham King authored Nov 21, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  f05f7629
- fix: missing tool calling usage (#4516) · 179ee38b
  Ayush Agarwal authored Nov 20, 2025
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  179ee38b
20 Nov, 2025 1 commit
- feat: Add metrics endpoint to kserve gRPC service (#4400) · 5fe0476e
  Tanmay Verma authored Nov 20, 2025
  
  5fe0476e
19 Nov, 2025 1 commit
- feat: nvext field to OpenAI APIs and add worker_id reporting (vLLM) (#4372) · 263f99dc
  Zhongxuan (Daniel) Wang authored Nov 18, 2025
```
Signed-off-by: Zhongxuan Wang <daniewang@nvidia.com>
```
  263f99dc
18 Nov, 2025 2 commits

feat: Support toolcall parser for DeepSeek V3 and R1 (#4253) · 164b0c29

tangcy98 authored Nov 19, 2025


Signed-off-by: zhangzhang <tangchenyu@xiaohongshu.com>
Co-authored-by: zhangzhang <tangchenyu@xiaohongshu.com>
Co-authored-by: Ayush Agarwal <ayushag@nvidia.com>

164b0c29

feat: support prompt_tokens_details in usage (#4239) · be9d6b2b
Vladislav Nosivskoy authored Nov 19, 2025
```
Signed-off-by: Vladislav Nosivskoy <vladnosiv@gmail.com>
```
be9d6b2b

17 Nov, 2025 2 commits
- refactor: centralize environment variable constants (#4083) · 0e77d344
  Keiven C authored Nov 17, 2025
```
Signed-off-by: Keiven Chang <keivenchang@users.noreply.github.com>
Co-authored-by: Keiven Chang <keivenchang@users.noreply.github.com>
```
  0e77d344
- feat: Command line flag to set request plane mode: tcp, http or nats (#4365) · 886506c1
  Graham King authored Nov 17, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  886506c1
13 Nov, 2025 1 commit
- chore: better error handling in prefill router (#4286) · ce833983
  Yan Ru Pei authored Nov 13, 2025
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  ce833983
11 Nov, 2025 1 commit
- chore: Remove static mode (#4235) · e1af3af6
  Graham King authored Nov 11, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  e1af3af6
10 Nov, 2025 2 commits
- refactor: consolidate Observability files (e.g. OTEL docker-compose, md files) (#4173) · afccc9d4
  Keiven C authored Nov 10, 2025
```
Signed-off-by: Keiven Chang <keivenchang@users.noreply.github.com>
Co-authored-by: Keiven Chang <keivenchang@users.noreply.github.com>
```
  afccc9d4
- refactor: Make the Runtime and DistributedRuntime fields private (#4193) · cf630bf7
  Graham King authored Nov 10, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  cf630bf7
08 Nov, 2025 3 commits
- fix: refactor to use service discovery (#4092) · 09b26bf6
  mohammedabdulwahhab authored Nov 08, 2025
```
Signed-off-by: mohammedabdulwahhab <furkhan324@berkeley.edu>
```
  09b26bf6
- fix: no more multiple finish reasons in stream (#4154) · 04f7579b
  Ayush Agarwal authored Nov 07, 2025
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  04f7579b
- feat: Add support for skip_special_tokens parameter in v1/completions and... · 441473c3
  Ryan McCormick authored Nov 07, 2025
```
feat: Add support for skip_special_tokens parameter in v1/completions and v1/chat/completions endpoints (#4175)
```
  441473c3
07 Nov, 2025 2 commits
- feat: enable HTTP completion endpoint to accept arrays of prompts and generate... · 93ada899
  zhongdaor-nv authored Nov 07, 2025
```
feat: enable HTTP completion endpoint to accept arrays of prompts and generate multiple completions per prompt (#3953)
Signed-off-by: zhongdaor <zhongdaor@nvidia.com>
```
  93ada899
- feat(keyvalue): Filesystem backed KeyValueStore (#4138) · 794c0a44
  Graham King authored Nov 07, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  794c0a44
06 Nov, 2025 1 commit
- feat: Reject unsupported parameters with 400 Bad Request (`/v1/completions`) (#4140) · 3d036fc4
  KrishnanPrash authored Nov 05, 2025
```
Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
```
  3d036fc4
05 Nov, 2025 1 commit
- fix: slipping <tool_calls_end> tool call parsing for DeepSeek v3.1 (#3995) · 7cfac6b8
  GuanLuo authored Nov 05, 2025
```
Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>
```
  7cfac6b8
04 Nov, 2025 1 commit
- feat: Image decoder in the frontend (#3971) · ae4b08ac
  milesial authored Nov 04, 2025
```
Signed-off-by: Alexandre Milesi <milesial@users.noreply.github.com>
```
  ae4b08ac
03 Nov, 2025 1 commit
- feat: Reject unsupported parameters with 400 Bad Request (#4021) · c837b5ba
  KrishnanPrash authored Nov 03, 2025
```
Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
```
  c837b5ba
31 Oct, 2025 1 commit
- feat: Media HTTP fetching and b64 decoding (#3967) · e30a3054
  milesial authored Oct 31, 2025
```
Signed-off-by: Alexandre Milesi <milesial@users.noreply.github.com>
```
  e30a3054
30 Oct, 2025 1 commit

feat: allow Triton model config specification in TensorModelConfig (#3874) · eb3a486d

GuanLuo authored Oct 31, 2025


Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>
Signed-off-by: GuanLuo <41310872+GuanLuo@users.noreply.github.com>

eb3a486d

29 Oct, 2025 1 commit
- chore: add usage field to non-streaming responses by default (#3922) · 3fe5653b
  Ayush Agarwal authored Oct 29, 2025
```
Signed-off-by: ayushag <ayushag@nvidia.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>
```
  3fe5653b
28 Oct, 2025 1 commit
- chore(runtime): Do not expose etcd lease ID (#3915) · c78b5901
  Graham King authored Oct 28, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  c78b5901
27 Oct, 2025 2 commits
- feat: Media URL passthrough in OAI preprocessor (#3733) · a79122c6
  milesial authored Oct 27, 2025
```
Signed-off-by: Alexandre Milesi <30204471+milesial@users.noreply.github.com>
```
  a79122c6
- chore(discovery): Use Store interface instead of etcd (#3887) · 5a0d710b
  Graham King authored Oct 27, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  5a0d710b
24 Oct, 2025 2 commits

feat: Extend TensorRequest and TensorResponse to contains extra parameters (#3761) · 90caf3ea

KrishnanPrash authored Oct 24, 2025

Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
Signed-off-by: KrishnanPrash <140860868+KrishnanPrash@users.noreply.github.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

90caf3ea

feat: add nats sink for audit logging (#3732) · cb0a3449
ryan-lempka authored Oct 23, 2025
```
Signed-off-by: Ryan Lempka <rlempka@nvidia.com>
```
cb0a3449

23 Oct, 2025 1 commit
- chore: Use KeyValueStoreManager instead of etcd::Client (#3822) · 7731b024
  Graham King authored Oct 23, 2025
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  7731b024
22 Oct, 2025 1 commit

fix: enhance gRPC frontend to return output in raw content field for Triton... · d68e4b8a

GuanLuo authored Oct 22, 2025

fix: enhance gRPC frontend to return output in raw content field for Triton client compatibility (#3600)
Signed-off-by: Guan Luo <gluo@nvidia.com>
Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>

d68e4b8a