SAVRN
Search Contact SAVRN

Open-weight model

visualrl-qwen3vl4b-ppo1-contrastsink-v2-r2-fix

by Zeyuan Yang Miiche/visualrl-qwen3vl4b-ppo1-contrastsink-v2-r2-fix

visualrl-qwen3vl4b-ppo1-contrastsink-v2-r2-fix is an open-weight model from Zeyuan Yang. Its published files total 110.5 GB.

Parameters
Context
Weights109.6 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-contrastsink-v2-r2-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
fc50c5abcdb73fc873b116bdb12ebe679c8baf30
First published
2026-09-19
Last updated
2026-09-20

Files and Weights

51 files, 110.5 GB in total. The weights are 26 files totalling 109.6 GB in pt.

Weights26 files · 109.6 GB
Configuration12 files · 9.5 KB
Tokenizer8 files · 31.8 MB
Other4 files · 866.4 MB
Repository1 file · 1.9 KB
Every file
FileTypeSizeSHA-256
global_step_20/actor/extra_state_world_size_4_rank_0.ptWeights15.1 KB c0c48b71a62b
global_step_20/actor/extra_state_world_size_4_rank_1.ptWeights15.1 KB c9692b1bb86f
global_step_20/actor/extra_state_world_size_4_rank_2.ptWeights15.1 KB 818eda81ec84
global_step_20/actor/extra_state_world_size_4_rank_3.ptWeights15.1 KB 452c67a00e1a
global_step_20/actor/model_world_size_4_rank_0.ptWeights4.8 GB 9388a2b0a4f2
global_step_20/actor/model_world_size_4_rank_1.ptWeights4.8 GB 1e008fc1b52e
global_step_20/actor/model_world_size_4_rank_2.ptWeights4.8 GB fbaf343a90a4
global_step_20/actor/model_world_size_4_rank_3.ptWeights4.8 GB 183dd9bcb25d
global_step_20/actor/optim_world_size_4_rank_0.ptWeights8.9 GB 0828436f84a8
global_step_20/actor/optim_world_size_4_rank_1.ptWeights8.9 GB 4a44c3e74e4a
global_step_20/actor/optim_world_size_4_rank_2.ptWeights8.9 GB 922a6f553fde
global_step_20/actor/optim_world_size_4_rank_3.ptWeights8.9 GB 64da7dbb9e28
global_step_20/dataloader.ptWeights2.0 KB 6f6c0b328df1
global_step_30/actor/extra_state_world_size_4_rank_0.ptWeights15.1 KB 5e9142b4ef60
global_step_30/actor/extra_state_world_size_4_rank_1.ptWeights15.1 KB cb7abe2522b3
global_step_30/actor/extra_state_world_size_4_rank_2.ptWeights15.1 KB 2f0e51628e80
global_step_30/actor/extra_state_world_size_4_rank_3.ptWeights15.1 KB f1b2d07a21c7
global_step_30/actor/model_world_size_4_rank_0.ptWeights4.8 GB af718a87cb54
global_step_30/actor/model_world_size_4_rank_1.ptWeights4.8 GB 035e597277d0
global_step_30/actor/model_world_size_4_rank_2.ptWeights4.8 GB 7f5f37f9ef40
global_step_30/actor/model_world_size_4_rank_3.ptWeights4.8 GB f7cf4031070e
global_step_30/actor/optim_world_size_4_rank_0.ptWeights8.9 GB b458408fb739
global_step_30/actor/optim_world_size_4_rank_1.ptWeights8.9 GB f6a0ced5f0e1
global_step_30/actor/optim_world_size_4_rank_2.ptWeights8.9 GB 0d2179a015f9
global_step_30/actor/optim_world_size_4_rank_3.ptWeights8.9 GB a86784380e85
global_step_30/dataloader.ptWeights2.0 KB b555ff47858c
global_step_20/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_20/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_20/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_20/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_20/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_20/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_30/actor/huggingface/added_tokens.jsonConfiguration707 B
global_step_30/actor/huggingface/config.jsonConfiguration1.6 KB
global_step_30/actor/huggingface/generation_config.jsonConfiguration213 B
global_step_30/actor/huggingface/preprocessor_config.jsonConfiguration782 B
global_step_30/actor/huggingface/special_tokens_map.jsonConfiguration613 B
global_step_30/actor/huggingface/video_preprocessor_config.jsonConfiguration817 B
global_step_20/actor/huggingface/chat_template.jinjaOther5.3 KB
global_step_30/actor/huggingface/chat_template.jinjaOther5.3 KB
logs/eval-train-q3vl4b_ppo1_contrastsink_v2_r2.logOther866.3 MB 100d95dc31a1
train_logs/train_metrics_q3vl4b_ppo1_contrastsink_v2_r2.jsonlOther60.5 KB
.gitattributesRepository1.9 KB
global_step_20/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_20/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_20/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_20/actor/huggingface/vocab.jsonTokenizer2.8 MB
global_step_30/actor/huggingface/merges.txtTokenizer1.7 MB
global_step_30/actor/huggingface/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_30/actor/huggingface/tokenizer_config.jsonTokenizer5.5 KB
global_step_30/actor/huggingface/vocab.jsonTokenizer2.8 MB

License and Download

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

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

Memory Requirements

PrecisionWeights in memory
As published109.6 GB

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