ehristoforu/qwen2.5-7b-upscaled
Partial fine-tune (fewer layers) vs Qwen/Qwen2.5-7B-Instruct. Qwen/Qwen2.5-7B-Instruct matches part of ehristoforu/qwen2.5-7b-upscaled (339 of its 555 tensors; the rest of ehristoforu/qwen2.5-7b-upscaled has no counterpart). On the shared part: Partial change: 24% of the weights changed (attention, mlp, norms), in 8 of 28 layers.
Tensors changed: Embeddings 0/1, Norms 16/57, MLP 24/84, Attention 56/196, lm_head 0/1
2 derivatives checked
- mradermacher/qwen2.5-7b-upscaled-i1-GGUF: Quantization
- mradermacher/qwen2.5-7b-upscaled-GGUF: Quantization