ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
unsloth/studio/backend/utils
History
Exact
Ayushman 304b8eca7a
 fix: match qwen3-thinking double-newline in train_on_responses_only response pattern (#6926) 
* fix: match qwen3-thinking chat template double-newline in response pattern

The Qwen3-thinking chat template generates `<think>\n\n` (double newline)
after the think tag, but `train_on_responses_only` was looking for
`<think>\n` (single newline).

`\n\n` is token 271 while `\n` is token 198 -- different tokens, so the
pattern match in `train_on_responses_only` fails, masking ALL tokens and
dropping 100% of training samples.

Update the response pattern from `<think>\n` to `<think>\n\n` to match
what the actual qwen3-thinking template generates.

Fixes #6919

* fix qwen3 thinking response marker

---------

Co-authored-by: Ayushman Paul <ayushman@HP>
Co-authored-by: Etherll <61019402+Etherll@users.noreply.github.com>
2026-07-07 21:28:18 +03:00
..
datasets fix: match qwen3-thinking double-newline in train_on_responses_only response pattern (#6926) 2026-07-07 21:28:18 +03:00
hardware Studio: multi-select export formats, portable FP8/INT8, GGUF LoRA, and source parity (#6767) 2026-07-03 08:25:10 -07:00
inference Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00
models (feat) Add project names to studio training runs (#6512) 2026-06-29 16:06:36 +02:00
paths Fix Studio custom folders on Linux external drives (#6799) 2026-07-03 19:10:04 +01:00
security Studio: persistent per-user trust_remote_code approval cache (#6551) 2026-06-22 05:12:49 -07:00
.gitkeep root studio folder 2026-02-02 09:13:49 +00:00
__init__.py Final cleanup 2026-03-12 18:28:04 +00:00
_studio_release_build.py Add Studio web update banner and release version display (#5308) 2026-05-11 18:24:01 +04:00
api_errors.py Studio: shareable per-checkpoint preview links (#6486) 2026-06-24 06:31:53 -07:00
cache_cleanup.py Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00
client_ip.py Studio: require signed capability tokens for /p preview links (#6666) 2026-06-25 21:40:48 -07:00
cpu_threads.py Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00
downsample.py Formatting: ruff line-length 100, kwarg-spacing passes, drop blank after short local imports (#6079) 2026-06-08 04:24:13 -07:00
embedding_model_settings.py Studio: customizable RAG embedding model with HF search, settings tab reorganization (#6800) 2026-07-02 05:26:33 -07:00
helper_precache_settings.py Studio: make Helper LLM startup pre-cache opt in (#6113) 2026-06-09 15:28:34 +02:00
hf_xet_fallback.py Auto Xet to HTTP download fallback in from_pretrained; share Studio's fallback via unsloth_zoo (#6638) 2026-07-06 05:13:25 -07:00
host_policy.py Studio: allow --secure with --api-only (headless secure API server) and add --api-only to unsloth studio run (#6591) 2026-06-23 05:44:56 -07:00
lifespan_shutdown.py Studio: make lifespan shutdown resilient to a dead default executor (#6307) 2026-06-15 22:51:46 -07:00
llama_cpp_freshness.py Studio: stop the llama.cpp update banner flickering and show the download size (#6338) 2026-06-17 21:28:54 -07:00
llama_cpp_update.py [Studio] Add --with-llama-cpp-dir installer flag to reuse a local llama.cpp (#6472) 2026-07-02 22:11:20 +01:00
mlx_repair.py Studio: exclude mlx-lm 0.31.3 (broke gemma4/qwen3_5 QK-norm load on Apple Silicon) (#6803) 2026-07-06 19:40:06 -07:00
native_path_leases.py Reap Studio child processes when the parent dies abnormally (#6425) 2026-06-18 05:51:22 -07:00
node_runtime.py Studio: use an isolated Node.js for the frontend build instead of replacing the system Node/npm (#6533) 2026-06-21 21:17:29 -07:00
openai_auto_switch_settings.py Studio: opt-in OpenAI /v1 model auto-switch and idle keep-warm (#6392) 2026-07-01 06:42:23 -07:00
personalization_settings.py studio: persist personalization (profile, avatar, theme) server-side (#6516) 2026-06-22 04:09:48 -07:00
preview_rate_limit.py Studio: require signed capability tokens for /p preview links (#6666) 2026-06-25 21:40:48 -07:00
preview_sharing_settings.py Studio: require signed capability tokens for /p preview links (#6666) 2026-06-25 21:40:48 -07:00
preview_token.py Studio: require signed capability tokens for /p preview links (#6666) 2026-06-25 21:40:48 -07:00
process_lifetime.py Reap Studio child processes when the parent dies abnormally (#6425) 2026-06-18 05:51:22 -07:00
ssm_runtime.py Auto-install SSM kernels (causal-conv1d, mamba-ssm) for inference loads (#6535) 2026-06-22 04:48:29 -07:00
studio_version.py Studio: llama.cpp update banner redesign, About tab license info, UI polish (#6196) 2026-06-11 09:27:34 -07:00
subprocess_compat.py Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00
training_runs.py (feat) Add project names to studio training runs (#6512) 2026-06-29 16:06:36 +02:00
transformers_version.py Studio: multi-select export formats, portable FP8/INT8, GGUF LoRA, and source parity (#6767) 2026-07-03 08:25:10 -07:00
update_status.py Studio: make code comments and docstrings more succinct (#6029) 2026-06-08 23:07:28 -07:00
upload_limits.py Formatting: ruff line-length 100, kwarg-spacing passes, drop blank after short local imports (#6079) 2026-06-08 04:24:13 -07:00
utils.py Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00
uv_path_safety.py Make _uv_safe_path space-safe on macOS/Linux (#6503) (#6534) 2026-06-24 04:02:24 -07:00
wheel_utils.py Reduce and tighten code comments and docstrings repo-wide (#6095) 2026-06-08 23:09:51 -07:00