chenyitian-shanshu/SIRL-COPT32B

Full fine-tune vs Qwen/Qwen2.5-32B-Instruct. Fine-tune: 95% of the weights changed, across 64 of 64 layers.

Tensors changed: Embeddings 0/1, Norms 120/129, MLP 192/192, Attention 447/448, lm_head 0/1

2 derivatives checked

• mradermacher/SIRL-COPT32B-i1-GGUF: Quantization
• mradermacher/SIRL-COPT32B-GGUF: Quantization

On the Hub · Downloads on Model Pulse