Open-weight model
visualrl-qwen3vl4b-ppo1-4bx-c02-clean
by Zeyuan Yang Miiche/visualrl-qwen3vl4b-ppo1-4bx-c02-clean
visualrl-qwen3vl4b-ppo1-4bx-c02-clean is an open-weight model from Zeyuan Yang. Its published files total 53.5 GB.
Parameters—
Context—
Weights38.6 GB
License—
AccessOpen weights
Monthly Downloads—
Model Card
The publisher has not written a card for this model.
Identity and Version
- Repository
- Miiche/visualrl-qwen3vl4b-ppo1-4bx-c02-clean
- Publisher
- Zeyuan Yang
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- Not stated by the source
- Languages
- Not stated by the source
- Revision
- 551072ddbe0a3b1e40b950100bc4690607557da6
- First published
- 2026-09-20
- Last updated
- 2026-09-22
Files and Weights
157 files, 53.5 GB in total. The weights are 9 files totalling 38.6 GB in safetensors.
Weights9 files · 38.6 GB
Configuration92 files · 171.8 MB
Tokenizer16 files · 63.5 MB
Other39 files · 14.6 GB
Repository1 file · 2.4 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| global_step_180_merged/model-00001-of-00003.safetensors | Weights | 4.3 GB | e494371f494d |
| global_step_180_merged/model-00002-of-00003.safetensors | Weights | 5.0 GB | f12509e3e57f |
| global_step_180_merged/model-00003-of-00003.safetensors | Weights | 390.1 MB | dd92172c5c95 |
| global_step_270_merged/model-00001-of-00002.safetensors | Weights | 5.0 GB | ed4a322d7f66 |
| global_step_270_merged/model-00002-of-00002.safetensors | Weights | 4.7 GB | 11f7e1596284 |
| global_step_80_merged/model-00001-of-00002.safetensors | Weights | 5.0 GB | 1dba99c8b08c |
| global_step_80_merged/model-00002-of-00002.safetensors | Weights | 4.7 GB | 732d7d69dcc9 |
| global_step_90_merged/model-00001-of-00002.safetensors | Weights | 5.0 GB | 8d92439ed460 |
| global_step_90_merged/model-00002-of-00002.safetensors | Weights | 4.7 GB | 949acda35558 |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/blink_output.json | Configuration | 6.8 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/blink_scores.json | Configuration | 3.2 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/chartqa_output.json | Configuration | 3.2 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/chartqa_scores.json | Configuration | 193 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/hallusionbench_output.json | Configuration | 1.9 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/hallusionbench_scores.json | Configuration | 492 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/logicvista_output.json | Configuration | 3.0 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/logicvista_scores.json | Configuration | 501 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathverse_output.json | Configuration | 17.0 MB | 67ae38c3bdc2 |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathverse_scores.json | Configuration | 2.3 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvision_output.json | Configuration | 3.1 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvision_scores.json | Configuration | 8.9 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvista_output.json | Configuration | 3.9 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvista_scores.json | Configuration | 7.5 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmbench_output.json | Configuration | 9.4 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmbench_scores.json | Configuration | 6.4 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mme_output.json | Configuration | 4.0 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mme_scores.json | Configuration | 2.0 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu_output.json | Configuration | 5.3 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu_pro_output.json | Configuration | 196.1 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu_pro_scores.json | Configuration | 110 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu_scores.json | Configuration | 2.7 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmstar_output.json | Configuration | 4.7 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmstar_scores.json | Configuration | 3.1 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmvet_output.json | Configuration | 249.1 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmvet_scores.json | Configuration | 505 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/perceptionbench_output.json | Configuration | 19.2 MB | a35fec46f79b |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/perceptionbench_scores.json | Configuration | 1.0 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/realworldqa_output.json | Configuration | 1.2 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/realworldqa_scores.json | Configuration | 95 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/scienceqa_output.json | Configuration | 3.4 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/scienceqa_scores.json | Configuration | 358 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/visnumbench_output.json | Configuration | 7.0 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/visnumbench_scores.json | Configuration | 949 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/blink_output.json | Configuration | 6.2 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/blink_scores.json | Configuration | 3.2 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/chartqa_output.json | Configuration | 3.0 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/chartqa_scores.json | Configuration | 193 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/hallusionbench_output.json | Configuration | 1.7 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/hallusionbench_scores.json | Configuration | 492 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/logicvista_output.json | Configuration | 2.5 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/logicvista_scores.json | Configuration | 500 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathverse_output.json | Configuration | 14.2 MB | 52ed6568ad8f |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathverse_scores.json | Configuration | 2.3 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathvista_output.json | Configuration | 3.6 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathvista_scores.json | Configuration | 7.5 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmbench_output.json | Configuration | 8.5 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmbench_scores.json | Configuration | 6.4 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mme_output.json | Configuration | 3.9 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mme_scores.json | Configuration | 2.0 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmmu_output.json | Configuration | 4.7 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmmu_scores.json | Configuration | 2.7 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmstar_output.json | Configuration | 4.3 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmstar_scores.json | Configuration | 3.1 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmvet_output.json | Configuration | 200.1 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmvet_scores.json | Configuration | 516 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/perceptionbench_output.json | Configuration | 15.3 MB | 9380b82c22be |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/perceptionbench_scores.json | Configuration | 1.0 KB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/realworldqa_output.json | Configuration | 1.1 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/realworldqa_scores.json | Configuration | 95 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/scienceqa_output.json | Configuration | 3.0 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/scienceqa_scores.json | Configuration | 359 B | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/visnumbench_output.json | Configuration | 5.6 MB | — |
| eval_results/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/visnumbench_scores.json | Configuration | 950 B | — |
| global_step_180_merged/added_tokens.json | Configuration | 707 B | — |
| global_step_180_merged/config.json | Configuration | 1.6 KB | — |
| global_step_180_merged/generation_config.json | Configuration | 147 B | — |
| global_step_180_merged/model.safetensors.index.json | Configuration | 64.8 KB | — |
| global_step_180_merged/preprocessor_config.json | Configuration | 782 B | — |
| global_step_180_merged/special_tokens_map.json | Configuration | 613 B | — |
| global_step_180_merged/video_preprocessor_config.json | Configuration | 817 B | — |
| global_step_270_merged/added_tokens.json | Configuration | 707 B | — |
| global_step_270_merged/config.json | Configuration | 1.6 KB | — |
| global_step_270_merged/generation_config.json | Configuration | 147 B | — |
| global_step_270_merged/model.safetensors.index.json | Configuration | 64.8 KB | — |
| global_step_270_merged/preprocessor_config.json | Configuration | 782 B | — |
| global_step_270_merged/special_tokens_map.json | Configuration | 613 B | — |
| global_step_270_merged/video_preprocessor_config.json | Configuration | 817 B | — |
| global_step_80_merged/added_tokens.json | Configuration | 707 B | — |
| global_step_80_merged/config.json | Configuration | 1.6 KB | — |
| global_step_80_merged/generation_config.json | Configuration | 147 B | — |
| global_step_80_merged/model.safetensors.index.json | Configuration | 64.8 KB | — |
| global_step_80_merged/preprocessor_config.json | Configuration | 782 B | — |
| global_step_80_merged/special_tokens_map.json | Configuration | 613 B | — |
| global_step_80_merged/video_preprocessor_config.json | Configuration | 817 B | — |
| global_step_90_merged/added_tokens.json | Configuration | 707 B | — |
| global_step_90_merged/config.json | Configuration | 1.6 KB | — |
| global_step_90_merged/generation_config.json | Configuration | 147 B | — |
| global_step_90_merged/model.safetensors.index.json | Configuration | 64.8 KB | — |
| global_step_90_merged/preprocessor_config.json | Configuration | 782 B | — |
| global_step_90_merged/special_tokens_map.json | Configuration | 613 B | — |
| global_step_90_merged/video_preprocessor_config.json | Configuration | 817 B | — |
| global_step_180_merged/chat_template.jinja | Other | 5.3 KB | — |
| global_step_270_merged/chat_template.jinja | Other | 5.3 KB | — |
| global_step_80_merged/chat_template.jinja | Other | 5.3 KB | — |
| global_step_90_merged/chat_template.jinja | Other | 5.3 KB | — |
| logs/eval-train-q3vl4b_ppo1_4bx_c02.log | Other | 14.6 GB | 73325f3d12d3 |
| logs/eval-visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm.log | Other | 2.2 MB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/blink.log | Other | 164.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/chartqa.log | Other | 107.7 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/hallusionbench.log | Other | 58.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/logicvista.log | Other | 47.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathverse.log | Other | 302.0 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvision.log | Other | 39.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mathvista.log | Other | 76.4 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmbench.log | Other | 189.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mme.log | Other | 118.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu.log | Other | 101.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmmu_pro.log | Other | 159.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmstar.log | Other | 90.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/mmvet.log | Other | 167.1 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/perceptionbench.log | Other | 252.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/realworldqa.log | Other | 57.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/scienceqa.log | Other | 100.1 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s80-vllm/visnumbench.log | Other | 166.4 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/blink.log | Other | 163.5 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/chartqa.log | Other | 108.4 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/hallusionbench.log | Other | 57.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/logicvista.log | Other | 53.3 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathverse.log | Other | 308.0 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mathvista.log | Other | 76.3 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmbench.log | Other | 188.8 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mme.log | Other | 118.6 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmmu.log | Other | 106.6 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmstar.log | Other | 94.4 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/mmvet.log | Other | 168.9 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/perceptionbench.log | Other | 268.2 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/realworldqa.log | Other | 59.1 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/scienceqa.log | Other | 100.0 KB | — |
| logs/visualrl-qwen3vl4b-ppo1-4bx-c02-clean-s90-vllm/visnumbench.log | Other | 171.9 KB | — |
| train_logs/train_metrics_q3vl4b_ppo1_4bx_c02.jsonl | Other | 623.7 KB | — |
| .gitattributes | Repository | 2.4 KB | — |
| global_step_180_merged/merges.txt | Tokenizer | 1.7 MB | — |
| global_step_180_merged/tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a |
| global_step_180_merged/tokenizer_config.json | Tokenizer | 5.5 KB | — |
| global_step_180_merged/vocab.json | Tokenizer | 2.8 MB | — |
| global_step_270_merged/merges.txt | Tokenizer | 1.7 MB | — |
| global_step_270_merged/tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a |
| global_step_270_merged/tokenizer_config.json | Tokenizer | 5.5 KB | — |
| global_step_270_merged/vocab.json | Tokenizer | 2.8 MB | — |
| global_step_80_merged/merges.txt | Tokenizer | 1.7 MB | — |
| global_step_80_merged/tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a |
| global_step_80_merged/tokenizer_config.json | Tokenizer | 5.5 KB | — |
| global_step_80_merged/vocab.json | Tokenizer | 2.8 MB | — |
| global_step_90_merged/merges.txt | Tokenizer | 1.7 MB | — |
| global_step_90_merged/tokenizer.json | Tokenizer | 11.4 MB | aeb13307a71a |
| global_step_90_merged/tokenizer_config.json | Tokenizer | 5.5 KB | — |
| global_step_90_merged/vocab.json | Tokenizer | 2.8 MB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 38.6 GB
Download from Zeyuan Yang
Released by Zeyuan Yang through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 38.6 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.