feat(webui): add seed controls for reproducibility

[kmishra1204]

What does this PR do?

Fixes #10581.

Adds explicit seed controls in the WebUI so users can reproduce generation, training, and evaluation/prediction runs without discovering the JSON extra_args escape hatch.

Changes:

• Adds a chat generation seed field, with -1 preserving the current random behavior.
• Passes generation seeds through the WebUI chatter path into HF, vLLM, and SGLang inference engines.
• Adds explicit training and evaluation seed fields in the WebUI and includes them in generated CLI args.
• Adds locale labels for the new controls.

Validation:

• uvx ruff check src\llamafactory\webui\components\chatbot.py src\llamafactory\webui\components\train.py src\llamafactory\webui\components\eval.py src\llamafactory\webui\chatter.py src\llamafactory\webui\runner.py src\llamafactory\chat\hf_engine.py src\llamafactory\chat\vllm_engine.py src\llamafactory\chat\sglang_engine.py src\llamafactory\webui\locales.py
• git diff --check
• py -m py_compile src\llamafactory\webui\components\chatbot.py src\llamafactory\webui\components\train.py src\llamafactory\webui\components\eval.py src\llamafactory\webui\chatter.py src\llamafactory\webui\runner.py src\llamafactory\chat\hf_engine.py src\llamafactory\chat\vllm_engine.py src\llamafactory\chat\sglang_engine.py src\llamafactory\webui\locales.py

Full runtime generation/training tests were not run locally because this environment does not have the heavyweight ML runtime dependencies installed, and I avoided pulling them for this focused WebUI/argument-wiring change.

Before submitting

• Did you read the contributor guideline?
• Did you write any new necessary tests? Not applicable for this WebUI/argument-wiring change; targeted lint, whitespace, and syntax checks are included above.

[gemini-code-assist]

Code Review

This pull request adds support for configuring random seeds across the training, evaluation, and chat interfaces in the WebUI, as well as in the Hugging Face, SGLang, and vLLM chat engines. The review feedback points out potential crash risks in the WebUI runner when parsing the 'train_seed' and 'eval_seed' textboxes, as calling int() directly on empty or non-numeric inputs will raise a ValueError. It is recommended to safely parse these inputs with a fallback default value.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

If the user clears the train_seed textbox or inputs an invalid/non-numeric value, calling int() directly on it will raise a ValueError and crash the WebUI runner thread.

To prevent this, we should safely parse the seed and fall back to the default seed of 42 if the input is empty or invalid.

Suggested change

	seed=int(get("train.train_seed")),
	seed=int(get("train.train_seed").strip()) if get("train.train_seed") and get("train.train_seed").strip().isdigit() else 42,

[gemini-code-assist]

Similarly to train_seed, if the user clears the eval_seed textbox or inputs an invalid/non-numeric value, calling int() directly on it will raise a ValueError and crash the WebUI runner thread.

We should safely parse the seed and fall back to the default seed of 42 if the input is empty or invalid.

Suggested change

	seed=int(get("eval.eval_seed")),
	seed=int(get("eval.eval_seed").strip()) if get("eval.eval_seed") and get("eval.eval_seed").strip().isdigit() else 42,

[kmishra1204]

Addressed the seed parsing review in the latest push.

Change made:

• Added a small _parse_seed(...) helper in webui/runner.py so blank or non-numeric training/eval seed textbox values fall back to the default 42 instead of raising ValueError.
• Kept chat generation seed handling unchanged because that path uses a numeric Gradio control and already normalizes negative values to the current random behavior.

Validation rerun:

• uvx ruff check src\llamafactory\webui\components\chatbot.py src\llamafactory\webui\components\train.py src\llamafactory\webui\components\eval.py src\llamafactory\webui\chatter.py src\llamafactory\webui\runner.py src\llamafactory\chat\hf_engine.py src\llamafactory\chat\vllm_engine.py src\llamafactory\chat\sglang_engine.py src\llamafactory\webui\locales.py
• py -m py_compile src\llamafactory\webui\components\chatbot.py src\llamafactory\webui\components\train.py src\llamafactory\webui\components\eval.py src\llamafactory\webui\chatter.py src\llamafactory\webui\runner.py src\llamafactory\chat\hf_engine.py src\llamafactory\chat\vllm_engine.py src\llamafactory\chat\sglang_engine.py src\llamafactory\webui\locales.py
• git diff --check