Quantized Qwen Language Models
19 repos
Optimized and compressed versions of Alibaba's Qwen language models, primarily focusing on quantization techniques (8-bit, Q4, Q5, Q6 formats) and the safetensors format for efficient model distribution and inference. The cluster contains multiple size variants of Qwen models (3B through 27B parameters) with various compression and distillation strategies applied, enabling deployment of capable language models on resource-constrained hardware. These repositories are primarily model weights and configurations rather than frameworks, representing a practical engineering focus on making state-of-the-art language models accessible across different computational budgets.
safetensors ·33
qwen3_5 ·32
quantized ·32
8-bit ·30
transformers ·28
conversational ·28
image-text-to-text ·28
endpoints_compatible ·28
AxionML ·24
modelopt ·24
