swiss-ai/Apertus-70B-Instruct-2509
Full fine-tune vs swiss-ai/Apertus-70B-2509. Fine-tune: 100% of the weights changed, across 80 of 80 layers.
Tensors changed: Embeddings 1/1, Norms 160/321, MLP 218/480, Attention 320/320, lm_head 1/1
12 derivatives checked
- unsloth/Apertus-70B-Instruct-2509-GGUF: Same weights, converted
- bartowski/swiss-ai_Apertus-70B-Instruct-2509-GGUF: Quantization
- tiny-random/apertus: Different architecture
- redponike/Apertus-70B-Instruct-2509-GGUF: Quantization
- EPFLiGHT/Apertus-70B-MeditronFO: Full fine-tune
- yujiepan/apertus-tiny-random: Different architecture
- giladgd/Apertus-70B-Instruct-2509-GGUF: Same weights, converted
- ssc-dsai/gc-llm-apertus-70b-instruct-2509: LoRA-style
- RedHatAI/Apertus-70B-Instruct-2509-FP8-dynamic: Quantization
- DevQuasar/swiss-ai.Apertus-70B-Instruct-2509-GGUF: Quantization
- RedHatAI/Apertus-70B-Instruct-2509-quantized.w4a16: Quantization
- Silverfoe/70B-Vegatron-Plastic: Full fine-tune
1 other repos hold byte-for-byte identical weight files.