swapnillo/Bangla-AI-1.7B
LoRA-style vs Qwen/Qwen3-1.7B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, q_proj, v_proj changed, in 28 of 28 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/113, MLP 0/84, Attention 84/112
2 derivatives checked
- mradermacher/Bangla-AI-1.7B-i1-GGUF: Quantization
- mradermacher/Bangla-AI-1.7B-GGUF: Same weights, converted