internlm/OREAL-32B

Full fine-tune vs internlm/OREAL-32B-SFT. Fine-tune: 95% of the weights changed, across 64 of 64 layers.

Tensors changed: Embeddings 1/1, Norms 122/129, MLP 187/192, Attention 443/448, lm_head 0/1

3 derivatives checked

• bartowski/internlm_OREAL-32B-GGUF: Quantization
• mradermacher/OREAL-32B-i1-GGUF: Quantization
• mradermacher/OREAL-32B-GGUF: Quantization

On the Hub · Downloads on Model Pulse