ArRENCEAI/TinyLlama-1.1B-Chat-v1.0-OBLITERATED

Partial fine-tune vs TinyLlama/TinyLlama-1.1B-Chat-v1.0. Partial change: 22% of the weights changed (attention, lm_head, mlp), in 4 of 22 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 0/45, MLP 12/66, Attention 16/88

On the Hub · Downloads on Model Pulse