ConvexAI/Luminex-34B-v0.1
LoRA-style vs jondurbin/bagel-34b-v0.2. 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 60 of 60 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/121, MLP 180/180, Attention 240/240, lm_head 0/1
2 derivatives checked
- mradermacher/Luminex-34B-v0.1-i1-GGUF: Quantization
- mradermacher/Luminex-34B-v0.1-GGUF: Quantization