oopere/Llama-3.2-1B-pruned-20pct

Same weights, converted vs meta-llama/Llama-3.2-1B. Same weights stored in another dtype (BF16 → F16).

Tensors changed: Embeddings 0/1, Norms 0/33, Attention 0/64

8 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse