SAVRN
Search Contact SAVRN

Open-weight model

visualrl-qwen3vl4b-ppo1-4bx-c50-hires-fix

by Zeyuan Yang Miiche/visualrl-qwen3vl4b-ppo1-4bx-c50-hires-fix

visualrl-qwen3vl4b-ppo1-4bx-c50-hires-fix is an open-weight model from Zeyuan Yang. Its published files total 200.7 GB.

Parameters
Context
Weights183.7 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-c50-hires-fix
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
33a8c4334f381a39731d24af21f1e220dc3bc7c5
First published
2026-09-19
Last updated
2026-09-20

Files and Weights

91 files, 200.7 GB in total. The weights are 44 files totalling 183.7 GB in pt.

Weights44 files · 183.7 GB
Configuration24 files · 18.9 KB
Tokenizer16 files · 63.5 MB
Other6 files · 16.9 GB
Repository1 file · 2.8 KB
Every file
FileTypeSizeSHA-256
global_step_120/actor/extra_state_world_size_4_rank_0.ptWeights15.1 KB ab1737f2f11d
global_step_120/actor/extra_state_world_size_4_rank_1.ptWeights15.1 KB 38b562a1212a
global_step_120/actor/extra_state_world_size_4_rank_2.ptWeights15.1 KB 3a5ea28adae2
global_step_120/actor/extra_state_world_size_4_rank_3.ptWeights15.1 KB b463370c7130
global_step_120/actor/model_world_size_4_rank_0.ptWeights4.8 GB 723760bb82ea
global_step_120/actor/model_world_size_4_rank_1.ptWeights4.8 GB be4176cbee10
global_step_120/actor/model_world_size_4_rank_2.ptWeights4.8 GB eb4ce5f3da24
global_step_120/actor/model_world_size_4_rank_3.ptWeights4.8 GB 9e30504a6224
global_step_120/actor/optim_world_size_4_rank_0.ptWeights8.9 GB 45a4b4a4a6e3
global_step_120/actor/optim_world_size_4_rank_1.ptWeights8.9 GB 8c9e119da268
global_step_120/actor/optim_world_size_4_rank_2.ptWeights8.9 GB 33c962ddbdb6
global_step_120/actor/optim_world_size_4_rank_3.ptWeights8.9 GB 2bef0667310c
global_step_120/dataloader.ptWeights2.0 KB 83c916d84e91
global_step_130/actor/extra_state_world_size_4_rank_0.ptWeights15.1 KB 4d30ef564d8e
global_step_130/actor/extra_state_world_size_4_rank_1.ptWeights15.1 KB 2ec9fab91593
global_step_130/actor/extra_state_world_size_4_rank_2.ptWeights15.1 KB 07e643828a24
global_step_130/actor/extra_state_world_size_4_rank_3.ptWeights15.1 KB b8f6dffb2564
global_step_130/actor/model_world_size_4_rank_0.ptWeights4.8 GB 6324be6b2036
global_step_130/actor/model_world_size_4_rank_1.ptWeights4.8 GB f803a3e13b60
global_step_130/actor/model_world_size_4_rank_2.ptWeights4.8 GB fe96cfb5569b
global_step_130/actor/model_world_size_4_rank_3.ptWeights4.8 GB 904e3f22e5ef
global_step_130/actor/optim_world_size_4_rank_0.ptWeights8.9 GB 89550321b542
global_step_130/actor/optim_world_size_4_rank_1.ptWeights8.9 GB 29f15c66ba6b
global_step_130/actor/optim_world_size_4_rank_2.ptWeights8.9 GB 8ead42c70505
global_step_130/actor/optim_world_size_4_rank_3.ptWeights8.9 GB 154096db27b8
global_step_130/dataloader.ptWeights2.0 KB 0c17a6757df4
global_step_90/actor/extra_state_world_size_4_rank_0.ptWeights15.1 KB 51d9abb1fd31
global_step_90/actor/extra_state_world_size_4_rank_1.ptWeights15.1 KB 31da6f58d250
global_step_90/actor/extra_state_world_size_4_rank_2.ptWeights15.1 KB b1a67b6aaf81
global_step_90/actor/extra_state_world_size_4_rank_3.ptWeights15.1 KB ed84249b2b36
global_step_90/actor/model_world_size_4_rank_0.ptWeights4.8 GB 493f0abfbc4b
global_step_90/actor/model_world_size_4_rank_1.ptWeights4.8 GB 0a335cc6cc76
global_step_90/actor/model_world_size_4_rank_2.ptWeights4.8 GB 8e77ee93b811
global_step_90/actor/model_world_size_4_rank_3.ptWeights4.8 GB 4072c230ed33
global_step_90/actor/optim_world_size_4_rank_0.ptWeights8.9 GB e890e6ec89f6
global_step_90/actor/optim_world_size_4_rank_1.ptWeights8.9 GB 0bc6f5e3cf8d
global_step_90/actor/optim_world_size_4_rank_2.ptWeights8.9 GB 61a18e864255
global_step_90/actor/optim_world_size_4_rank_3.ptWeights8.9 GB 96576b0427ff
global_step_90/dataloader.ptWeights2.0 KB 9f371cfbfa32
global_step_90_highlight/actor/model_world_size_4_rank_0.ptWeights4.8 GB 493f0abfbc4b
global_step_90_highlight/actor/model_world_size_4_rank_1.ptWeights4.8 GB 0a335cc6cc76
global_step_90_highlight/actor/model_world_size_4_rank_2.ptWeights4.8 GB 8e77ee93b811
global_step_90_highlight/actor/model_world_size_4_rank_3.ptWeights4.8 GB 4072c230ed33
global_step_90_highlight/dataloader.ptWeights2.0 KB 9f371cfbfa32
global_step_120/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_120/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_120/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_120/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_120/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_120/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_130/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_130/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_130/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_130/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_130/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_130/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_90/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_90/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_90/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_90/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_90/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_90/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_90_highlight/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_90_highlight/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_90_highlight/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_90_highlight/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_90_highlight/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_90_highlight/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_120/actor/huggingface/chat_template.jinjaOther5.3 KB
global_step_130/actor/huggingface/chat_template.jinjaOther5.3 KB
global_step_90/actor/huggingface/chat_template.jinjaOther5.3 KB
global_step_90_highlight/actor/huggingface/chat_template.jinjaOther5.3 KB
logs/eval-train-q3vl4b_ppo1_4bx_c50_hires.logOther16.9 GB 11a26f4d8586
train_logs/train_metrics_q3vl4b_ppo1_4bx_c50_hires.jsonlOther262.1 KB
.gitattributesRepository2.8 KB
global_step_120/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_120/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_120/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_120/actor/huggingface/vocab.jsonTokenizer2.8 MB
global_step_130/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_130/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_130/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_130/actor/huggingface/vocab.jsonTokenizer2.8 MB
global_step_90/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_90/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_90/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_90/actor/huggingface/vocab.jsonTokenizer2.8 MB
global_step_90_highlight/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_90_highlight/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_90_highlight/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_90_highlight/actor/huggingface/vocab.jsonTokenizer2.8 MB

License and Download

License
Not stated by the source
Access
Open weights, no gate
Download size
183.7 GB
Download from Zeyuan Yang

Released by Zeyuan Yang through its official repository on Hugging Face.

Memory Requirements

PrecisionWeights in memory
As published183.7 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.