Default branch

main 22493242a3 · Studio: Don't re-prompt finished answers in the tool loop (#7505) · Updated 2026-07-29 11:38:17 +02:00

ZIP  TAR.GZ

Branches

dh/recover-4183-hermes-chat-template 5281ddae64 · Fix Hermes chat template validation error · Updated 2026-03-12 10:16:07 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-4179-lora-chat-template 661194623e · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 10:16:03 +01:00    ceos	3036 3	Compare	ZIP  TAR.GZ
dh/recover-4177-modernbert-num-labels a454c212d8 · Using a loop to handle the sequence classification parameters. · Updated 2026-03-12 10:14:18 +01:00    ceos	3036 3	Compare	ZIP  TAR.GZ
dh/recover-4090-idefics3 831aea71cf · Add Idefics3 support (Granite Docling VLM) · Updated 2026-03-12 10:10:02 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-4077-ddp-vlm-tiled-mlp e32db1e1a1 · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 10:09:17 +01:00    ceos	3036 4	Compare	ZIP  TAR.GZ
dh/recover-3891-cggr 458f2a93ef · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 09:57:42 +01:00    ceos	3036 3	Compare	ZIP  TAR.GZ
dh/recover-3816-vllm-revision f003507878 · fix: propagate revision to vLLM fast_inference [skip pre-commit.ci] · Updated 2026-03-12 09:43:46 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3804-zero-stride-fix 3d64e11cc7 · fix: handle zero-strided tensors in fast_rope_embedding (#3781) [skip pre-commit.ci] · Updated 2026-03-12 09:42:47 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3153-seq2seq 718f673439 · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 09:19:58 +01:00    ceos	3036 2	Compare	ZIP  TAR.GZ
dh/recover-3691-ruff-version 98b1eab4a8 · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 09:19:31 +01:00    ceos	3036 2	Compare	ZIP  TAR.GZ
dh/recover-3469-filelocks a7037b6785 · Filelocks · Updated 2026-03-12 09:12:34 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3660-trl-native-datasets eb18f02673 · Fix: Add support for TRL native dataset formats · Updated 2026-03-12 09:10:18 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3241-lr-strict-mode 01cdf0f974 · feat: removing hard limits on learning rate · Updated 2026-03-12 09:07:43 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3341-reference-docs e558b71df9 · [pre-commit.ci] auto fixes from pre-commit.com hooks · Updated 2026-03-12 08:51:43 +01:00    ceos	3036 2	Compare	ZIP  TAR.GZ
dh/recover-3234-eval-steps-warning 16f5b21adf · refactor(trainer): add warning for ignored eval_steps · Updated 2026-03-12 08:49:22 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3793-revision-chat-template ae38c3639d · fix: add revision parameter to FastBaseModel in vision.py · Updated 2026-03-12 08:37:43 +01:00    ceos	3036 3	Compare	ZIP  TAR.GZ
dh/recover-3436-qwen25-coder f35e027ece · Add Qwen2.5 Coder model support to registry and chat templates · Updated 2026-03-12 08:33:45 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-3425-gpt-oss-bf16 345fe0373a · add gptoss BF16 to models mapper · Updated 2026-03-12 08:32:19 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-1225-load-checkpoint-add-new-tokens-rebased ef98930175 · fix/load-checkpoint-add-new-tokens · Updated 2026-03-12 07:59:40 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ
dh/recover-974-cuda-visible-devices-rebased 08abf2f8a8 · Fix for multi gpu setup training with a single GPU. · Updated 2026-03-12 07:59:40 +01:00    ceos	3036 1	Compare	ZIP  TAR.GZ