Popular repositories Loading

1. french-llm-from-scratch french-llm-from-scratch Public

   Training a French GPT model from scratch: 260M params, 130M tokens, with fine-tuning on conversations. Full pipeline with dashboard, checkpoints & LR scheduling.

   Python 22 1

2. BindCraft-dockeready BindCraft-dockeready Public

   Forked from martinpacesa/BindCraft

   docker ready version of the famous user friendly and accurate binder design pipeline

   Python

3. Open_Turboquant Open_Turboquant Public

   High-performance LLM Quantization & KV Cache Compression Engine. Featuring PolarQuant technology, universal model patching, and 4-bit KV caching for extreme memory efficiency.

   Python

4. HyperQwen HyperQwen Public

   Forked from syv-ai/HyperQwen

   Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent…

   Python