Skip to content

Actions: Roblox/FAI-RL

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
33 workflow runs
33 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
v0.2.4 Upload Python Package #30: Release v0.2.4 published by yw-roblox
41s
41s
v0.2.3 Upload Python Package #29: Release v0.2.3 published by yw-roblox
41s
41s
v0.2.2 Upload Python Package #28: Release v0.2.2 published by yw-roblox
56s
56s
v0.2.1 Upload Python Package #27: Release v0.2.1 published by yw-roblox
51s
51s
v0.2.0 Upload Python Package #26: Release v0.2.0 published by yw-roblox
44s
44s
Gemma-4-31B recipes + VLM LoRA vision-tower exclusion, single-GPU DataParallel fix Upload Python Package #25: Release v0.1.43 published by yw-roblox
34s
34s
feat(llm-finetune): inference system/user chat prompt split Upload Python Package #24: Release v0.1.42 published by yw-roblox
34s
34s
feat(sft): add completion-only loss via split user/assistant prompts Upload Python Package #23: Release v0.1.41 published by yw-roblox
33s
33s
support embedded-image parquet datasets Upload Python Package #22: Release v0.1.40 published by yw-roblox
34s
34s
support multiple image columns for VLM inference Upload Python Package #21: Release v0.1.39 published by yw-roblox
40s
40s
support multiple image and text columns Upload Python Package #20: Release v0.1.38 published by yw-roblox
38s
38s
fix(config): remove dangling value_model_name reference left from PPO removal Upload Python Package #19: Release v0.1.37 published by yw-roblox
35s
35s
treat failed model uploads as failures instead of reporting success Upload Python Package #18: Release v0.1.36 published by yw-roblox
43s
43s
0.1.35 Upload Python Package #17: Release v0.1.35_2 published by yw-roblox
38s
38s
let DeepSpeed derive train_batch_size for ZeRO-3 zero.Init Upload Python Package #16: Release v0.1.35 published by yw-roblox
32s
32s
Add the version to the log Upload Python Package #15: Release v0.1.34 published by yw-roblox
44s
44s
resolve DeepSpeed 'auto' batch fields before ZeRO-3 zero Upload Python Package #14: Release v0.1.33 published by yw-roblox
32s
32s
register ZeRO-3 HfDeepSpeedConfig before model load Upload Python Package #13: Release v0.1.32 published by yw-roblox
39s
39s
Add multimodal (vision-language) SFT Upload Python Package #12: Release v0.1.31 published by yw-roblox
50s
50s
Upload the inference/eval result to S3 Upload Python Package #11: Release v0.1.30 published by yw-roblox
42s
42s
inference from S3 model Upload Python Package #10: Release v0.1.29 published by yw-roblox
35s
35s
enabled ignore_mismatched_sizes Upload Python Package #9: Release v0.1.27 published by yw-roblox
39s
39s
fix sync command Upload Python Package #8: Release v0.1.26 published by yw-roblox
46s
46s
Upgrade transformer version to the latest Upload Python Package #7: Release v0.1.25 published by yw-roblox
38s
38s
Roll back transformer Upload Python Package #6: Release v0.1.24 published by yw-roblox
48s
48s