Fetching metadata from the HF Docker repository...
Apr 12, 2025 08:14 add attention mask 08:06 Clean model description 08:00 pin torch to 2.4.0 07:59 add sentencepiece tokenzier 07:55 update requirements 07:54 enable zerogpu with decorator 07:43 Apply ZeroGPU 06:35 add apt.txt 06:10 adjust requirements 04:58 adjust requirements.txt 04:56 specify gpu build for llama cpp python 04:03 Update README.md 04:02 fix readme 04:00 switch to gradio version for stability reason Apr 11, 2025 16:45 add taiwan tinyllama 14:39 Improve responsiveness by asynchronously retrieving web search context - Run web search lookup in a background thread to decouple it from prompt construction.
- Add a 2-second timeout using thread.join(timeout=2) to prevent delaying response generation.
- Fallback to using only the system prompt and user query if web search results aren't retrieved in time.
- Update sidebar UI to show the fetched context or a "No context found" message.
- Retain current streaming and chat history functionalities for model inference. 14:11 update requirements \ 14:04 specify wheel of llama cpp python 13:55 openblas build of llam acpp filed 13:23 specify to build llama cpp with openblas 08:05 add llama 3.2 taiwan 3b 07:58 add minicpm3 4b 07:52 increase xt length to max 07:50 remove all moe 07:40 remove qwen 1.5 moe 07:38 adjust title style 07:37 use another version of qwen 1.5 moe 07:33 add Qwen1.5-MoE 07:15 Qwen2.5-MOE-6x1.5B 06:22 remove under 3b models 06:04 Add model caching 03:36 UI/UX Improvement 01:46 reset timeout timer once a new token is generated Apr 10, 2025 15:39 open web search settgins to user 15:23 add 2 more models 15:08 apply new settings on duckduck search 15:02 tune llama paramters 14:41 increase max_chars_per_result to 600 14:40 increase max results to 6 for better web search 14:14 increase ctx lenght to 2k 13:50 increase timeout to 5min 13:29 Code simplification 12:49 Enable speculattive decoding 10:58 fix role disorder error in history 08:34 Add internet search feature 08:00 fix reasonning model's thought process display 07:44 bugfix 06:27 fix for multi-turn conv. 06:11 bugfix for think tag handling 06:05 support reasoning tag 
