trt19/gemma-300m-legal-ft1

Partial fine-tune vs google/embeddinggemma-300m. Partial change: 34% of the weights changed (attention, mlp, norms), in 24 of 24 layers.

Tensors changed: Embeddings 0/1, Norms 145/145, MLP 72/72, Attention 96/96

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

On the Hub · Downloads on Model Pulse