Skip to page content
SAVRN
Search Plan a project

Open-weight model

visualrl-qwen3vl2b-ppo1-4ax-c10-clean

by Zeyuan Yang Miiche/visualrl-qwen3vl2b-ppo1-4ax-c10-clean

visualrl-qwen3vl2b-ppo1-4ax-c10-clean is an open-weight model from Zeyuan Yang. Its published files total 37.7 GB.

Parameters—
Context—
Weights14.6 GB
License—
AccessOpen weights
Monthly Downloads—

Model Card

The publisher has not written a card for this model.

Identity and Version

Repository
Miiche/visualrl-qwen3vl2b-ppo1-4ax-c10-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
8f62f255a822ff86721b14fbfcbe6c0e6045592c
First published
2026-09-21
Last updated
2026-09-22

Files and Weights

176 files, 37.7 GB in total. The weights are 3 files totalling 14.6 GB in safetensors.

Weights3 files · 14.6 GB
Configuration108 files · 329.6 MB
Tokenizer12 files · 47.6 MB
Other52 files · 22.7 GB
Repository1 file · 3.1 KB
Every file
FileTypeSizeSHA-256
global_step_180_merged/model.safetensorsWeights4.9 GB f1973e94cfa6
global_step_270_merged/model.safetensorsWeights4.9 GB f0fb3d50663a
global_step_90_merged/model.safetensorsWeights4.9 GB d6094dcf144d
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/blink_output.jsonConfiguration8.3 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/blink_scores.jsonConfiguration3.2 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/chartqa_output.jsonConfiguration2.0 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/chartqa_scores.jsonConfiguration193 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/hallusionbench_output.jsonConfiguration1.7 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/hallusionbench_scores.jsonConfiguration492 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/logicvista_output.jsonConfiguration3.7 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/logicvista_scores.jsonConfiguration492 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathverse_output.jsonConfiguration21.5 MB b19d0af92259
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathverse_scores.jsonConfiguration2.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvision_output.jsonConfiguration4.0 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvision_scores.jsonConfiguration8.9 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvista_output.jsonConfiguration5.1 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvista_scores.jsonConfiguration7.5 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmbench_output.jsonConfiguration11.3 MB 02f25200f645
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmbench_scores.jsonConfiguration6.4 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mme_output.jsonConfiguration2.5 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mme_scores.jsonConfiguration2.0 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu_output.jsonConfiguration5.4 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu_pro_output.jsonConfiguration196.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu_pro_scores.jsonConfiguration108 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu_scores.jsonConfiguration2.7 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmstar_output.jsonConfiguration5.6 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmstar_scores.jsonConfiguration3.1 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmvet_output.jsonConfiguration276.8 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmvet_scores.jsonConfiguration503 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/perceptionbench_output.jsonConfiguration18.8 MB f6740d4c4888
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/perceptionbench_scores.jsonConfiguration1.0 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/realworldqa_output.jsonConfiguration1.1 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/realworldqa_scores.jsonConfiguration95 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/scienceqa_output.jsonConfiguration5.2 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/scienceqa_scores.jsonConfiguration359 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/visnumbench_output.jsonConfiguration10.5 MB 9097f87ef884
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/visnumbench_scores.jsonConfiguration952 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/blink_output.jsonConfiguration9.7 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/blink_scores.jsonConfiguration3.2 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/chartqa_output.jsonConfiguration3.5 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/chartqa_scores.jsonConfiguration193 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/logicvista_output.jsonConfiguration4.8 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/logicvista_scores.jsonConfiguration503 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmbench_output.jsonConfiguration13.8 MB 2d40d8743366
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmbench_scores.jsonConfiguration6.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mme_output.jsonConfiguration3.3 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mme_scores.jsonConfiguration2.0 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmstar_output.jsonConfiguration7.0 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmstar_scores.jsonConfiguration3.1 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmvet_output.jsonConfiguration346.2 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmvet_scores.jsonConfiguration477 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/perceptionbench_output.jsonConfiguration24.4 MB 7915a280c4a9
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/perceptionbench_scores.jsonConfiguration1.1 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/realworldqa_output.jsonConfiguration2.1 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/realworldqa_scores.jsonConfiguration95 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/scienceqa_output.jsonConfiguration7.5 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/scienceqa_scores.jsonConfiguration358 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/visnumbench_output.jsonConfiguration12.0 MB 4c64aa3ee969
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/visnumbench_scores.jsonConfiguration953 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/blink_output.jsonConfiguration10.4 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/blink_scores.jsonConfiguration3.2 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/chartqa_output.jsonConfiguration2.2 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/chartqa_scores.jsonConfiguration193 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/hallusionbench_output.jsonConfiguration2.5 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/hallusionbench_scores.jsonConfiguration491 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/logicvista_output.jsonConfiguration4.3 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/logicvista_scores.jsonConfiguration504 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathverse_output.jsonConfiguration26.1 MB 403c5020b287
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathverse_scores.jsonConfiguration2.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvision_output.jsonConfiguration4.6 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvision_scores.jsonConfiguration8.7 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvista_output.jsonConfiguration5.6 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvista_scores.jsonConfiguration7.4 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmbench_output.jsonConfiguration13.8 MB eb4f3358694a
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmbench_scores.jsonConfiguration6.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mme_output.jsonConfiguration3.1 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mme_scores.jsonConfiguration2.0 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu_output.jsonConfiguration6.5 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu_pro_output.jsonConfiguration196.3 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu_pro_scores.jsonConfiguration108 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu_scores.jsonConfiguration2.7 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmstar_output.jsonConfiguration6.8 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmstar_scores.jsonConfiguration3.1 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmvet_output.jsonConfiguration253.8 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmvet_scores.jsonConfiguration527 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/perceptionbench_output.jsonConfiguration26.7 MB 16e1f94313fd
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/perceptionbench_scores.jsonConfiguration1.0 KB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/realworldqa_output.jsonConfiguration1.8 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/realworldqa_scores.jsonConfiguration95 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/scienceqa_output.jsonConfiguration6.2 MB —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/scienceqa_scores.jsonConfiguration358 B —
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/visnumbench_output.jsonConfiguration12.9 MB a23ed3b8f281
eval_results/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/visnumbench_scores.jsonConfiguration953 B —
global_step_180_merged/added_tokens.jsonConfiguration707 B —
global_step_180_merged/config.jsonConfiguration1.6 KB —
global_step_180_merged/generation_config.jsonConfiguration147 B —
global_step_180_merged/preprocessor_config.jsonConfiguration782 B —
global_step_180_merged/special_tokens_map.jsonConfiguration613 B —
global_step_180_merged/video_preprocessor_config.jsonConfiguration817 B —
global_step_270_merged/added_tokens.jsonConfiguration707 B —
global_step_270_merged/config.jsonConfiguration1.6 KB —
global_step_270_merged/generation_config.jsonConfiguration147 B —
global_step_270_merged/preprocessor_config.jsonConfiguration782 B —
global_step_270_merged/special_tokens_map.jsonConfiguration613 B —
global_step_270_merged/video_preprocessor_config.jsonConfiguration817 B —
global_step_90_merged/added_tokens.jsonConfiguration707 B —
global_step_90_merged/config.jsonConfiguration1.6 KB —
global_step_90_merged/generation_config.jsonConfiguration147 B —
global_step_90_merged/preprocessor_config.jsonConfiguration782 B —
global_step_90_merged/special_tokens_map.jsonConfiguration613 B —
global_step_90_merged/video_preprocessor_config.jsonConfiguration817 B —
global_step_180_merged/chat_template.jinjaOther5.3 KB —
global_step_270_merged/chat_template.jinjaOther5.3 KB —
global_step_90_merged/chat_template.jinjaOther5.3 KB —
logs/eval-train-q3vl2b_ppo1_4ax_c10.logOther22.7 GB fa0e23192a0d
logs/eval-visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm.logOther2.2 MB —
logs/eval-visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm.logOther2.1 MB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/blink.logOther186.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/chartqa.logOther133.1 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/hallusionbench.logOther66.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/logicvista.logOther41.7 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathverse.logOther257.8 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvision.logOther29.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mathvista.logOther82.6 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmbench.logOther241.2 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mme.logOther141.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu.logOther100.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmmu_pro.logOther157.6 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmstar.logOther101.0 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/mmvet.logOther53.2 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/perceptionbench.logOther266.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/realworldqa.logOther69.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/scienceqa.logOther123.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s180-vllm/visnumbench.logOther116.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/blink.logOther188.8 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/chartqa.logOther136.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/logicvista.logOther35.9 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmbench.logOther238.1 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mme.logOther145.0 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmstar.logOther100.6 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/mmvet.logOther54.8 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/perceptionbench.logOther257.9 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/realworldqa.logOther76.2 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/scienceqa.logOther130.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s270-vllm/visnumbench.logOther105.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/blink.logOther189.2 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/chartqa.logOther140.6 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/hallusionbench.logOther69.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/logicvista.logOther39.1 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathverse.logOther243.0 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvision.logOther25.4 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mathvista.logOther82.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmbench.logOther243.1 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mme.logOther149.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu.logOther98.7 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmmu_pro.logOther147.0 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmstar.logOther101.9 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/mmvet.logOther53.3 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/perceptionbench.logOther252.1 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/realworldqa.logOther72.0 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/scienceqa.logOther124.5 KB —
logs/visualrl-qwen3vl2b-ppo1-4ax-c10-clean-s90-vllm/visnumbench.logOther101.7 KB —
train_logs/train_metrics_q3vl2b_ppo1_4ax_c10.jsonlOther545.0 KB —
.gitattributesRepository3.1 KB —
global_step_180_merged/merges.txtTokenizer1.7 MB —
global_step_180_merged/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_180_merged/tokenizer_config.jsonTokenizer5.5 KB —
global_step_180_merged/vocab.jsonTokenizer2.8 MB —
global_step_270_merged/merges.txtTokenizer1.7 MB —
global_step_270_merged/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_270_merged/tokenizer_config.jsonTokenizer5.5 KB —
global_step_270_merged/vocab.jsonTokenizer2.8 MB —
global_step_90_merged/merges.txtTokenizer1.7 MB —
global_step_90_merged/tokenizer.jsonTokenizer11.4 MB aeb13307a71a
global_step_90_merged/tokenizer_config.jsonTokenizer5.5 KB —
global_step_90_merged/vocab.jsonTokenizer2.8 MB —

License and Download

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

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

Memory Requirements

PrecisionWeights in memory
As published14.6 GB

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