arcee-ai/Llama-3.1-SuperNova-Lite
Partial fine-tune vs meta-llama/Llama-3.1-8B-Instruct. Partial change: 68% of the weights changed (attention, mlp, norms), in 32 of 32 layers.
Tensors changed: Embeddings 0/1, Norms 14/65, MLP 87/96, Attention 32/128, lm_head 0/1
19 derivatives checked
- bartowski/Llama-3.1-SuperNova-Lite-GGUF: Same weights, converted
- legraphista/Llama-3.1-SuperNova-Lite-IMat-GGUF: Same weights, converted
- mradermacher/Llama-3.1-SuperNova-Lite-i1-GGUF: Quantization
- mradermacher/Llama-3.1-SuperNova-Lite-GGUF: Same weights, converted
- backyardai/Llama-3.1-SuperNova-Lite-GGUF: Quantization
- QuantFactory/Baldur-8B-GGUF: Full fine-tune
- featherless-ai-quants/arcee-ai-Llama-3.1-SuperNova-Lite-GGUF: Quantization
- DevQuasar-5/Llama-3.1-SuperNova-Lite-GGUF: Quantization
- tensorblock/Llama-3.1-SuperNova-Lite-GGUF: Quantization
- AuraIndustries/Aura-8B: Full fine-tune
- grimjim/Llama-3.1-SuperNova-Lite-lorabilterated-8B: Partial fine-tune
- meditsolutions/Llama-3.1-MedIT-SUN-8B: Identical copy
- SE6446/Llama-3.1-SuperNova-Lite-Reflection-V1.0: LoRA-style
- grimjim/Llama3.1-SuperNovaLite-HuatuoSkywork-o1-8B: Full fine-tune
- Youlln/1PARAMMYL-8B-ModelStock: Full fine-tune
- bunnycore/Llama-3.1-8B-OmniMatrix: Full fine-tune
- bunnycore/HyperLLama3.1-8b-Nova: Full fine-tune
- Edens-Gate/CAI-Supernova-r1: Full fine-tune
- Edens-Gate/CAI-Supernova-r2: Full fine-tune
1 other repos hold byte-for-byte identical weight files.