open-machine/Llama-3.1-8B-FlashNorm-test

Partial fine-tune vs meta-llama/Llama-3.1-8B. Partial change: 63% of the weights changed (attention, lm_head, mlp, norms), in 32 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 65/65, MLP 64/96, Attention 96/128, lm_head 1/1

2 derivatives checked

• mradermacher/Llama-3.1-8B-FlashNorm-i1-GGUF: Quantization
• mradermacher/Llama-3.1-8B-FlashNorm-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse