OpenPipe/ART

Issues

• Training freezes at 0% "train" after "gather" stage is complete

  #346 opened 3 months ago by Aranxtonel
  14

• AttributeError: 'LLM' object has no attribute 'engine_step' with Unsloth and vLLM

  #411 opened 2 months ago by Gjw2333
  0

• Issue with json_schema in MCP RL

  #410 opened 2 months ago by wm19999
  0

• Always Block when train

  #401 opened 2 months ago by johnson7788
  1

• [Feature Request] Azure OpenAI Support

  #409 opened 2 months ago by jgbradley1
  0

• RuntimeError: torch.cuda.MemPool doesn't currently support expandable_segments during vLLM model initialization

  #405 opened 2 months ago by abhinav262666
  0

• typerror: unexpected kwarg `enable_prefix_caching`

  #387 opened 2 months ago by ExpressGradient
  3

• [Feature Request] Add integration to smolagents

  #398 opened 2 months ago by FacerAin
  2

• [Question] About 2048 scoring

  #394 opened 2 months ago by katopz
  2

• Question on Training Efficiency and API Cost Compared to Standard RL

  #399 opened 2 months ago by l-besiege-l
  1

• Training crashes after sometime

  #386 opened 2 months ago by aravindc26
  1

• xml tools use

  #400 opened 2 months ago by wm19999
  0

• About appropriate groups_per_step parameter setting given training and validation dataset volume

  #397 opened 2 months ago by llv22
  0

• Allow running multiple `art.LocalBackend` sessions on different processes

  #396 opened 2 months ago by giladfrid009
  0

• Training stuck

  #377 opened 2 months ago by linkailong555-del
  5

• train process hang out after a few steps

  #395 opened 2 months ago by h66840
  1

• Add support for SFT through ART directly

  #348 opened 3 months ago by saum7800
  1

• question in art_e/rollout.py

  #391 opened 2 months ago by aliasGitee
  0

• Bug: art.rewards.ruler function and module collision

  #339 opened 3 months ago by giladfrid009
  3

• Ran into a few walls with LangGraph Integration Example

  #366 opened 2 months ago by ryaneggz
  4

• Feature: Dr. GRPO Support

  #364 opened 2 months ago by giladfrid009
  0

• training stuck

  #315 opened 3 months ago by oraby8
  3

• Add @art.rollout decorator to gather trajectories

  #347 opened 3 months ago by arcticfly
  6

• try mcp-14b-alpha-001, it block by below error, how can i know where the trigger it?

  #343 opened 3 months ago by johnson7788
  1

• Support `exclude` parameter for SkyPilotBackend `_experimental_pull_from_s3`

  #350 opened 3 months ago by arcticfly
  0

• MCP with grpo

  #345 opened 3 months ago by OliverLeeXZ
  1

• Unable to reach OpenAI-compatible server

  #329 opened 3 months ago by Jim2016713
  3

• RuntimeError: CUDA error: device-side assert triggered when running 2048.ipynb

  #344 opened 3 months ago by watemailunpi
  1

• Suggestion: Potential major performance improvement for `gather_trajectory_groups` with `after_each`

  #342 opened 3 months ago by giladfrid009
  0

• Only works with AsyncOpenAI client, not synchronous OpenAI client

  #341 opened 3 months ago by Aranxtonel
  2

• Add error message if log_probs not found and allow_training_without_logprobs=False

  #336 opened 3 months ago by mattiasutancykeln
  2

• Upgrade VLLM dependency to support transformers 4.54.0+ (AIMv2 registration conflict)

  #272 opened 3 months ago by ExpressGradient
  3

• About asynchronous generation and training like AReal

  #337 opened 3 months ago by llv22
  2

• Cannot train the tau-bench datasets on qwen3 model

  #270 opened 3 months ago by aongwachi1
  9

• Gradient reliability with sample-by-sample vs batch processing

  #285 opened 3 months ago by zfflxx
  2

• bug:ValueError: 'LoRARequest' object has no attribute 'lora_tensors'

  #313 opened 3 months ago by linkailong555-del
  17

• Could I add a translated README.md file in Chinese(Simplified)?

  #271 opened 3 months ago by 8086CJ
  1

• Feature request - load best checkpoint

  #299 opened 3 months ago by giladfrid009
  2

• closing Local backend - bug

  #288 opened 3 months ago by giladfrid009
  2

• Bug - No checkpoint metrics recorded during training

  #300 opened 3 months ago by giladfrid009
  1

• EngineArgs clarification

  #280 opened 3 months ago by giladfrid009
  3

• prompt opt without training

  #284 opened 3 months ago by genji970
  1

• Timeout issue

  #281 opened 3 months ago by giladfrid009
  1

• Importing client-side code errors on mac

  #291 opened 3 months ago by arcticfly
  0

• langgraph create_react_agent

  #275 opened 3 months ago by austinmw
  4

• ImportError: cannot import name 'ConstantLengthDataset' from 'trl.trainer.utils'

  #276 opened 3 months ago by JonesAndrew
  1

• GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning - Research Discussion

  #274 opened 3 months ago by jmanhype
  1

• Any video tutorial of this project ?

  #269 opened 3 months ago by Greatz08
  1

• `load_trajectories` requires backend dependencies

  #266 opened 4 months ago by arcticfly
  0

• enforce_eager=True not being respected in test scripts

  #262 opened 4 months ago by corbt
  2