jinaai/jina-embeddings-v5-omni-nano-text-matching
LoRA-style vs jinaai/jina-embeddings-v5-omni-nano. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 12 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: other 0/5, MLP 36/212, Norms 0/203, Attention 48/320, Embeddings 0/4
1 derivatives checked
- jinaai/jina-embeddings-v5-omni-nano-text-matching-GGUF: Same weights, converted (fewer layers)