chenjoya/LiveCC-7B-Instruct

Partial fine-tune vs Qwen/Qwen2-VL-7B. Partial change: 85% of the weights changed (attention, lm_head, mlp, norms), in 28 of 32 layers.

Tensors changed: Embeddings 0/2, Norms 56/187, MLP 84/216, Attention 196/324, lm_head 1/1

2 derivatives checked

• mradermacher/LiveCC-7B-Instruct-i1-GGUF: Quantization (fewer layers)
• mradermacher/LiveCC-7B-Instruct-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse