Fetching metadata from the HF Docker repository...
Jun 15 13:46 refactor: implement _summary_list_field function to streamline handling of summary fields in coach summary; enhance payload processing and validation for string and list types 13:36 refactor: streamline evidence generation in coach summary; enhance JSON output contract and improve tokenization handling for local chat 13:15 refactor: add functions for JSON payload extraction and validation in coach summary; streamline summary key handling and improve error reporting 13:07 refactor: introduce compact representation functions for metrics, issues, and knowledge cards; enhance summary evidence generation and improve error handling in JSON extraction 12:56 refactor: enhance JSON response handling in coach summary generation; update response format instructions and improve error reporting for invalid JSON outputs 12:34 chore: remove Linux-specific dependencies from pyproject.toml and requirements.txt; update README to clarify usage of Nemotron implementation for local model inference 12:07 chore: add ninja-build and build-essential to package requirements; update README and documentation to reflect changes in model configuration and local Transformers usage 08:57 chore: update base model to NVIDIA-Nemotron-3-Nano-4B-BF16 across configurations and documentation; adjust max sequence length to 2048; upgrade GitHub Actions cache to v5 08:15 refactor: enhance _encode_bt709_output and _encode_bt709_command for improved audio stream handling and code clarity 08:07 refactor: remove sample_pose_cache module and streamline pose processing in pipeline Jun 14 23:24 refactor: change default delegate to CPU in MediaPipePoseBackend and update tests accordingly NLag
23:05 refactor: update artifact URLs and improve router device handling for better classification performance NLag
22:49 feat: add support for local Transformers inference on Hugging Face Spaces with opt-in configuration NLag
22:40 refactor(mediapipe): simplify GPU delegate preference logic and remove unused CUDA visibility check NLag
21:58 refactor(mediapipe): enhance MediaPipePoseBackend to prefer tasks on GPU and improve delegate handling NLag
21:41 docs: update default coach-summary model to build-small-hackathon/pozify-coach-summary1 and clarify fallback behavior in documentation 21:20 feat(coach): disable verifier by default 10:17 feat: add coach summary LoRA config and provider plumbing 10:17 feat: prepare filtered public fitness style corpus 10:16 feat: build Pozify-native coach summary SFT datasets 10:11 feat: expand coach retrieval with equipment and goal overlays 10:11 feat: ingest real exercise dataset into knowledge artifacts 10:00 feat: add exercise dataset to card-pack transformer 09:39 feat: load external coach knowledge packs 20:48 feat(slm): add llama.cpp coach summary provider 20:25 feat(pipeline): implement sample pose caching and enhance upload handling - Added a caching mechanism for sample pose sequences to improve performance and reduce redundant processing.
- Introduced a new function to sanitize uploaded file names for temporary storage.
- Updated the analysis pipeline to utilize cached pose sequences when available, enhancing efficiency.
- Refactored related tests to ensure proper functionality of caching and upload processes. 12:04 feat(coach): add local transformers provider Jun 11 21:36 fix(hf-spaces): install mediapipe native libraries 21:23 fix(hf-spaces): avoid pickling analysis progress callbacks 21:07 feat(pose)!: use coco17 3d router features BREAKING CHANGE: exercise router artifacts must be regenerated with the coco17_3d_v1 feature schema. 18:51 Implement grounded coach summary with HF SLM 15:47 feat(hf): use ZeroGPU for analysis 15:07 Set default exercise router repo NLag
14:55 perf(pose): read dense video frames sequentially 14:47 pioritize BiLSTM model NLag
14:34 Refactor(pipeline): improve code readability and update function calls for clarity 14:26 Merge branch 'main' into NLag/issue4 14:24 refactor(mediapipe): improve code readability and add CPU delegate option 14:21 feat(report): stream analysis progress 14:15 Enhance UnknownExercise class with debug data handling and add test for manual unknown exercise in pipeline NLag
14:03 Avoid OpenCV H264 writer fallback errors NLag
13:37 feat(report): add annotated issue clips 13:29 Merge remote issue 4 updates NLag
13:29 Add issue 10 training release assets NLag
12:44 Merge branch 'main' into NLag/issue4 12:36 fix(renderer): preserve video colors in annotations 12:21 refactor(exercises): route pipeline through exercise objects 11:42 Implement frame-level issue markers 09:39 Merge branch 'main' into NLag/issue4 08:39 refactor(exercises): introduce exercise strategies 
