pthinc/prettybird_bce_basic_vl_8b

LoRA-style (fewer layers) vs Qwen/Qwen2.5-VL-7B-Instruct. pthinc/prettybird_bce_basic_vl_8b matches part of Qwen/Qwen2.5-VL-7B-Instruct (339 of its 729 tensors; the rest of Qwen/Qwen2.5-VL-7B-Instruct has no counterpart). On the shared part: pthinc/prettybird_bce_basic_vl_8b is a GGUF (FP16) of different weights. 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 28 of 28 layers; norms, biases and embeddings are untouched.

Tensors changed: Embeddings 0/1, Norms 0/57, MLP 84/84, Attention 112/196, lm_head 0/1

On the Hub · Downloads on Model Pulse