EVA-UNIT-01/EVA-Qwen2.5-32B-v0.0
Partial fine-tune vs Qwen/Qwen2.5-32B. Partial change: 50% of the weights changed (attention, lm_head, mlp, norms), in 64 of 64 layers.
Tensors changed: Embeddings 0/1, Norms 64/129, MLP 96/192, Attention 224/448, lm_head 1/1
4 derivatives checked
- bartowski/EVA-Qwen2.5-32B-v0.0-GGUF: Same weights, converted
- mradermacher/EVA-Qwen2.5-32B-v0.0-i1-GGUF: Quantization
- mradermacher/EVA-Qwen2.5-32B-v0.0-GGUF: Quantization
- ParasiticRogue/EVA-Instruct-32B: Full fine-tune