nbeerbower/GigaPotato-Mistral-Nemo-23B
Partial fine-tune (fewer layers) vs nbeerbower/mistral-nemo-kartoffel-12B. nbeerbower/mistral-nemo-kartoffel-12B matches part of nbeerbower/GigaPotato-Mistral-Nemo-23B (363 of its 723 tensors; the rest of nbeerbower/GigaPotato-Mistral-Nemo-23B has no counterpart). On the shared part: Partial change: 87% of the weights changed (attention, mlp, norms), in 39 of 40 layers.
Tensors changed: Embeddings 0/1, Norms 78/81, MLP 117/120, Attention 156/160, lm_head 0/1
3 derivatives checked
- mradermacher/GigaPotato-Mistral-Nemo-23B-i1-GGUF: Quantization
- mradermacher/GigaPotato-Mistral-Nemo-23B-GGUF: Quantization
- nbeerbower/DoppelKartoffel-Mistral-Nemo-23B: LoRA-style