Qwen/Qwen2-VL-2B-Instruct
Partial fine-tune vs Qwen/Qwen2-VL-2B. Partial change: 71% of the weights changed (attention, embeddings, mlp, norms), in 28 of 32 layers.
Tensors changed: Embeddings 1/2, Norms 57/187, MLP 88/216, Attention 196/324
74 derivatives checked
- numind/NuExtract-2.0-2B: Full fine-tune
- jinaai/jina-reranker-m0: Full fine-tune
- Qwen/Qwen2-VL-2B-Instruct-AWQ: Quantization
- bartowski/Qwen2-VL-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- ggml-org/Qwen2-VL-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- Alibaba-NLP/gme-Qwen2-VL-2B-Instruct: Partial fine-tune
- unsloth/Qwen2-VL-2B-Instruct-bnb-4bit: Quantization
- unsloth/Qwen2-VL-2B-Instruct-unsloth-bnb-4bit: Quantization
- second-state/Qwen2-VL-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- unsloth/Qwen2-VL-2B-Instruct: Same weights, repacked
- mradermacher/Qwen2-VL-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- Mungert/Nanonets-OCR2-1.5B-exp-GGUF: Full fine-tune (fewer layers)
- prithivMLmods/Qwen2-VL-OCR-2B-Instruct: LoRA-style
- mobilint/Qwen2-VL-2B-Instruct: Full fine-tune (fewer layers)
- mradermacher/Qwen2-VL-2B-Instruct-i1-GGUF: Quantization (fewer layers)
- OpenSearch-AI/Ops-MM-embedding-v1-2B: LoRA-style
- imadityasarkar/satquery-phase2-vrsbench: Adapter
- Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int4: Quantization
- PollardWeights/FlyBrain-Pollard-VL-CNSv1: Quantization (fewer layers)
- lightonai/MonoQwen2-VL-v0.1: Adapter
- kulibinai/cadevolve-rl1: Full fine-tune
- JZPeterPan/MedVLM-R1: Full fine-tune
- microsoft/GUI-Actor-2B-Qwen2-VL: Full fine-tune
- AdaptLLM/remote-sensing-Qwen2-VL-2B-Instruct: Partial fine-tune
- gaianet/Qwen2-VL-2B-Instruct-GGUF: Same weights, converted (fewer layers)
- prithivMLmods/Qwen2-VL-2B-Abliterated-Caption-it: LoRA-style
- Nanthasit/sakthai-vision-7b: Not comparable yet
- asuglia/Qwen2-VL-2B-Instruct-Q4_K_M-GGUF: Quantization (fewer layers)
- nanonets/Nanonets-OCR2-1.5B-exp: Full fine-tune
- Qwen/Qwen2-VL-2B-Instruct-GPTQ-Int8: Quantization
13 other repos hold byte-for-byte identical weight files.