ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
933 commits 391 branches 58 tags 304 MiB
Commit graph

28 commits

Author SHA1 Message Date
Shine1i
3b1663b1e9 feat(recipe-studio, datasets): improve dataset handling and update metadata logic 2026-03-09 02:47:32 +01:00
Roland Tannous
1dab9f57b6 fix: validate pip exit codes for .venv_t5 installs in setup.ps1 2026-03-07 23:25:13 +00:00
Roland Tannous
ef9184c731 fix: prevent training hang on Windows by adding triton-windows support 2026-03-07 17:53:36 +00:00
Roland Tannous
29fa91be07 fix: bump transformers to 5.2.0 and pin huggingface_hub in setup.ps1 2026-03-07 09:12:12 +00:00
Roland Tannous
f8bd4303f7 feat: subprocess-based training for transformers version switching 2026-03-05 15:40:32 +00:00
Roland Tannous
e280e457d1 Move llama.cpp clone/build from in-tree to ~/.unsloth/llama.cpp 
- setup.sh: builds at ~/.unsloth/llama.cpp instead of ./llama.cpp
- setup.ps1: builds at %USERPROFILE%/.unsloth/llama.cpp
- inference llama_cpp.py: searches ~/.unsloth/ first, in-tree as legacy
- export.py: updated comments (unsloth-zoo handles path natively)
 2026-03-02 04:04:41 +00:00
Roland Tannous
674cc67d78 Tighten Python bounds to >= 3.11, < 3.14 (matching setup.sh), only auto-install if missing 2026-03-01 13:05:10 +00:00
Roland Tannous
d5644d2d0d Add Python 3.12 prerequisite check with auto-install via winget 2026-03-01 13:05:10 +00:00
Roland Tannous
0267ba0a18 Auto-enable Windows Long Paths via UAC elevation during setup 2026-03-01 13:05:10 +00:00
Roland Tannous
6536bfb33b Remove unused CMP0194 cmake policy (eliminates cmake warning) 2026-03-01 13:05:10 +00:00
Roland Tannous
453f423d22 Simplify: use winget OpenSSL.Dev instead of vcpkg for HTTPS support 2026-03-01 13:05:10 +00:00
Roland Tannous
2e102b683e Add vcpkg/curl[ssl] for HTTPS support in llama-server, enable LLAMA_CURL=ON 2026-03-01 13:05:10 +00:00
Roland Tannous
9eb0ff074b Add .venv/Scripts to User PATH so unsloth-studio works without activation 2026-03-01 13:05:10 +00:00
Roland Tannous
8e22b16bd8 Simplify completion banner: no venv activation needed 2026-03-01 13:05:10 +00:00
Roland Tannous
12867f701b Auto-add CUDA DLLs to PATH when launching llama-server on Windows 2026-03-01 13:05:10 +00:00
Roland Tannous
d79fe439ed Warn user to uninstall incompatible CUDA toolkit instead of failed side-by-side 2026-03-01 13:05:10 +00:00
Roland Tannous
b22c5b6ed8 Fallback: try descending CUDA versions if exact driver-max install fails 2026-03-01 13:05:10 +00:00
Roland Tannous
7b4d074857 Always persist compatible CUDA_PATH to User registry (overwrite stale values) 2026-03-01 13:05:10 +00:00
Roland Tannous
7576552717 Fix: scan side-by-side CUDA installs, pick compatible toolkit version 2026-03-01 13:05:10 +00:00
Roland Tannous
3521de7040 Build llama.cpp in-tree, auto-detect driver CUDA version for compatible toolkit 2026-03-01 13:05:10 +00:00
Roland Tannous
8d272ff8d5 Auto-detect driver CUDA version, install compatible toolkit instead of latest 2026-03-01 13:05:10 +00:00
Roland Tannous
bd7c17708b Set short TORCHINDUCTOR_CACHE_DIR to fix Windows MAX_PATH crash 2026-03-01 13:05:10 +00:00
Roland Tannous
e1cc5e61b1 Fix npm Invalid Version: delete package-lock.json, relax Node constraint 2026-03-01 13:05:10 +00:00
Roland Tannous
aba3d8e29b Enforce Node LTS (v20-v22), add npm error checking, clean node_modules 2026-03-01 13:05:10 +00:00
Roland Tannous
2dfe0abaa1 Fix npm stderr crash on Windows ErrorActionPreference 2026-03-01 13:05:10 +00:00
Roland Tannous
662a1eb9d5 Fix Windows frontend build, add setup.bat, ANSI colors, aliases 2026-03-01 13:05:10 +00:00
Roland Tannous
783f0caf5f add setup.bat 2026-03-01 13:05:10 +00:00
Roland Tannous
28bac1859a Extract shared install_python_stack.py for cross-platform setup 2026-03-01 13:05:10 +00:00