- Replaced `CookBookIcon` with `ChefHatIcon` in navbar for improved clarity.
- Added dark mode-specific gradient styles to recipe cards for better visual differentiation.
Remove os.chdir(save_directory) from export.py which was causing all of
unsloth-zoo's relative-path internals (check_llama_cpp, use_local_gguf,
_download_convert_hf_to_gguf) to resolve against the export directory
instead of the repo root. This caused llama.cpp to be cloned inside each
export dir and destroyed the repo root's llama-server build on cleanup.
Now passes absolute paths to save_pretrained_gguf so unsloth resolves
llama.cpp from the repo root where setup.sh already built it.
Also builds llama-quantize in setup.sh (needed by unsloth-zoo's export
pipeline) and symlinks it to llama.cpp root for check_llama_cpp().
- Replace datetime.UTC with datetime.timezone.utc in authentication.py and storage.py
- Fixes ImportError on Python versions < 3.11
- timezone.utc works on Python 3.9+
Resolves#237
Fixes two bugs:
1. Chat template tags (<|im_start|>, <|im_end|>) leaking into output
because /v1/completions treated them as literal text
2. Image hallucination because image_b64 was never passed to llama-server
Now llama-server handles chat templates natively and receives images
as OpenAI-format multimodal content parts for vision models.
Replace Python-side GGUF download with llama-server's native -hf flag for
HuggingFace repos. Add frontend variant picker so users can choose
quantization (Q4_K_M, Q8_0, BF16, etc.) with file sizes. Fix vision
detection via mmproj files instead of hardcoding is_vision=False.
* Fix FP8 model loading for BNB/16-bit: redirect to BF16 sibling
Models like Ministral-3-3B-Instruct-2512 ship with FP8 weights and an FP8
quantization_config in their config.json. Loading these with BNB 4-bit/8-bit
fails because BNB cannot quantize FP8 tensors. Loading with 16-bit also fails
because the FP8 quantization config has activation_scheme=static which is
unsupported by transformers' FineGrainedFP8Config.
When an FP8 model is detected and the user is not explicitly requesting FP8
loading, check if a BF16 sibling repo exists (model_name + "-BF16") and
redirect to it. This happens early in the loading flow before any quantization
config processing.
Also pass the modified model_config to auto_model.from_pretrained to avoid
transformers re-reading the original config from the model repo.
Tested with Ministral-3-3B in 4-bit and 16-bit modes. Both now load and
train correctly.
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Simplify FP8 condition and narrow exception handling
Simplify the load_in_fp8 check (works for bool and string values).
Narrow inner except to KeyError and add comment for outer except.
* Warn user when FP8 model has no BF16 sibling for redirect
Previously the except block silently fell through with `pass`,
so users would get a confusing BNB dtype error later. Now prints
a clear message explaining the FP8 situation and suggesting
load_in_fp8=True or uploading a BF16 version.
* Fix FP8 redirect state corruption and add fbgemm_fp8 support
- Fix state corruption: model_name was reassigned before
AutoConfig.from_pretrained, so if config fetch failed,
model_name pointed to BF16 repo while auto_config still
had FP8. Now only updates state after both checks succeed.
- Save original model_name so warning message is correct
even on failure.
- Handle fbgemm_fp8 quant method in addition to fp8.
* Extract FP8 redirect to shared _redirect_fp8_to_bf16() in _utils.py
Addresses reviewer feedback:
- Move FP8 redirect logic to a shared function callable from both
vision.py (FastBaseModel) and llama.py (FastLlamaModel)
- Raise RuntimeError instead of warning when BF16 sibling not found
- Add FP8 redirect to llama.py for text-only model loading path
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
* Add Ministral 3B/8B/14B mapper entries
Adds all 9 Ministral model variants to the mapper:
- Instruct (3B, 8B, 14B) with FP8 variant mappings
- Base (3B, 8B, 14B)
- Reasoning (3B, 8B, 14B)
This routes mistralai/Ministral-* to unsloth/Ministral-* repos
(BF16 weights), which also avoids the FP8 config issue for the
standard loading path through loader.py.
* Add FP8 mapper entries for Mistral-Small-3.2 and Magistral-Small-2509
---------
Co-authored-by: Ubuntu <ubuntu@ip-172-31-16-253.us-east-2.compute.internal>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
- Added constrained dependency files for single-env installations: `constraints.txt`, `data-designer.txt`, and `data-designer-deps.txt`.
- Implemented a `patch_metadata.py` script to resolve metadata conflicts between dependency versions.
- Updated `setup.sh` to integrate single-env setup, including dependency installation and metadata patching.
- Upgraded `fastmcp` and `websockets` versions in `extras.txt` for compatibility.
- Commented out unused "Start Tutorial" button in `data-recipes-page.tsx`.
- Added support for configuring markdown note block styles, including color and opacity.
- Enabled double-click on markdown notes to open their configuration dialog.
- Adjusted layout styles in markdown previews for better interaction control.
- Updated relevant payloads, types, and UI logic to support added styling features.
- Integrated multiple example notes in learning recipes for better visualization.
- Added "Markdown Note" block to allow users to add UI-only markdown notes to the canvas for documentation purposes.
- Integrated note creation, editing, and rendering in the `recipe-studio` UI, including markdown previews.
- Updated payload generation logic to omit markdown notes from backend payloads.
- Enhanced block types, definitions, and dialog support to include the new "Markdown Note" feature.
- Introduced "Multi-Turn Chat" recipe to generate structured user-assistant conversations with domain/topic-based goals and constraints.
- Added `conversation.json` with model configuration, sampling strategies, and LLM prompts.
- Updated UI nodes, layout, and graph rendering logic to support new recipe.
- Enhanced `recipe-studio` fit view logic to improve editor layout responsiveness.
- Added three new learning recipes: "Instruction from Answer," "PDF Grounded QA," and "Structured Outputs Jinja," with respective metadata and configuration.
- Integrated support for unstructured and structured input handling, including sampling strategies, prompt definitions, and model specifications.
- Enhanced JSON structure and UI nodes to facilitate better recipe visualization and execution.
- Introduced `layoutDirection` to control graph orientation ("LR" or "TB") and integrate into edges, nodes, and payloads.
- Enhanced handle management with new default, semantic, and data-specific mappings based on layout direction.
- Added handle normalization for consistent connections across layouts and semantic/data flows.
- Updated UI to reflect layout-aware positioning and semantic connections.
- Added handle normalization functions to standardize handle IDs across connections.
- Expanded UI for scorer options with real-time updates, input fields for values and descriptions, and support for adding/removing options.
- Updated graph node handles and their layout logic for better connection visualization.
- Stripped sensitive fields (e.g., `api_key`) from payloads during export.
- Introduced a new "Instruction from Answer" learning recipe with related metadata, payload integration, and UI updates.
- Enhanced badge display logic to include up to 3 badges with overflow indication for additional learning badges.
