Qwen/Qwen2.5-Math-7B
Full fine-tune vs Qwen/Qwen2.5-7B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 57/57, MLP 84/84, Attention 196/196, lm_head 1/1
48 derivatives checked
- Qwen/Qwen2.5-Math-7B-Instruct: Full fine-tune
- BayesRL/Llama3.1-IVON-SFT-8B: Different architecture
- unsloth/Qwen2.5-Math-7B: Same weights, repacked
- Mungert/OpenMath-Nemotron-7B-GGUF: Full fine-tune
- Mungert/Aryabhata-1.0-GGUF: Full fine-tune
- QuantFactory/Qwen2.5-Math-7B-Instruct-GGUF: Full fine-tune
- mradermacher/Qwen2.5-Math-7B-i1-GGUF: Quantization
- featherless-ai-quants/Qwen-Qwen2.5-Math-7B-GGUF: Quantization
- zbeeb/Qwen2.5-Math-7B-GRPO-Staleness-2: Full fine-tune
- nvidia/OpenMath-Nemotron-7B: Full fine-tune
- QuantFactory/BFS-Prover-GGUF: Full fine-tune
- ByteDance-Seed/BFS-Prover-V2-7B: Full fine-tune
- mradermacher/Qwen2.5-Math-7B-GGUF: Same weights, converted
- DevQuasar-8/Qwen.Qwen2.5-Math-7B-GGUF: Same weights, converted
- ByteDance-Seed/BFS-Prover-V1-7B: Full fine-tune
- sail/Qwen2.5-Math-7B-Oat-Zero: Full fine-tune
- nvidia/NFT-7B: Full fine-tune
- Dongwei/Qwen-2.5-7B_Base_Math_smalllr: Full fine-tune
- open-r1/Qwen2.5-Math-7B-RoPE-300k: Identical copy
- unsloth/Qwen2.5-Math-7B-bnb-4bit: Quantization
- BayesRL/Qwen2.5Math-IVON-SFT-7B: Full fine-tune
- zbeeb/Qwen2.5-Math-7B-GRPO-Staleness-4: Full fine-tune
- TIGER-Lab/Qwen2.5-Math-7B-CFT: Full fine-tune
- hkust-nlp/Qwen-2.5-Math-7B-SimpleRL-Zero: Full fine-tune
- Elliott/LUFFY-Qwen-Math-7B-Zero: Full fine-tune
- PhysicsWallahAI/Aryabhata-1.0: Full fine-tune
- Yuqian-Fu/SRFT-Qwen2.5-Math-7B: Full fine-tune
- Satori-reasoning/Satori-7B-Round2: Full fine-tune
- InfiX-ai/InfiAlign-Qwen-7B-DPO: Full fine-tune
- JingHaoZ/RLFR-Qwen2.5-Math-7B: Full fine-tune
7 other repos hold byte-for-byte identical weight files.