Pinned Loading
-
modelscope/ms-swift
modelscope/ms-swift PublicUse PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.8, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM5.3, Gemma4, Llava, P…
-
modelscope/twinkle
modelscope/twinkle PublicTwinkle✨: Training workbench to make your model glow.
-
modelscope/modelscope
modelscope/modelscope PublicModelScope: bring the notion of Model-as-a-Service to life.
-
modelscope/mcore-bridge
modelscope/mcore-bridge PublicMCore-Bridge: Providing Megatron-Core model definitions for state-of-the-art large models and making Megatron training as simple as Transformers — with support for 300+ large language models (Qwen3…
-
QwenLM/Qwen3
QwenLM/Qwen3 PublicQwen3 is the large language model series developed by Qwen team, Alibaba Cloud.
-
vllm-project/vllm
vllm-project/vllm PublicA high-throughput and memory-efficient inference and serving engine for LLMs
If the problem persists, check the GitHub status page or contact support.



