JoPmt/Trismal-Obrum-AlBe-7B-Base-Ties

Partial fine-tune vs Locutusque/OpenCerebrum-1.5-Mistral-7B-v0.2-beta. Partial change: 3% of the weights changed (attention, mlp, norms), in 7 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 1/65, MLP 3/96, Attention 6/128, lm_head 0/1

2 derivatives checked

• mradermacher/Trismal-Obrum-AlBe-7B-Base-Ties-GGUF: Same weights, converted
• JoPmt/Trismal-Obrum-v2-7B-Base-Ties: Full fine-tune

On the Hub · Downloads on Model Pulse