SirSahOl/Tev1-4B-experimental-chat-mlx-16bit

Partial fine-tune (fewer layers) vs togethercomputer/Tev1-4B-experimental. SirSahOl/Tev1-4B-experimental-chat-mlx-16bit matches part of togethercomputer/Tev1-4B-experimental (402 of its 738 tensors; the rest of togethercomputer/Tev1-4B-experimental has no counterpart). On the shared part: Partial change: 0% of the weights changed (norms), in 32 of 32 layers.

Tensors changed: Embeddings 0/1, MLP 0/96, Attention 0/200, Norms 81/105

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

On the Hub · Downloads on Model Pulse