BEE-spoke-data/zephyr-220m-sft-full

Partial fine-tune vs BEE-spoke-data/smol_llama-220M-openhermes. Partial change: 85% of the weights changed (attention, lm_head, mlp, norms), in 10 of 10 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 21/21, MLP 30/30, Attention 40/40

2 derivatives checked

• afrideva/zephyr-220m-sft-full-GGUF: Same weights, converted
• BEE-spoke-data/zephyr-220m-dpo-full: Full fine-tune

On the Hub · Downloads on Model Pulse