Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / allenai/open-instruct issues and pull requests

#555 - [WIP] removing incorrect date cutoffs

Pull Request - State: open - Opened by natolambert 5 days ago

#554 - Add train file support back to finetune

Pull Request - State: closed - Opened by hamishivi 6 days ago

#553 - Use dpo_tune_cache

Pull Request - State: closed - Opened by ljvmiranda921 6 days ago

#552 - Fix eval script

Pull Request - State: closed - Opened by hamishivi 7 days ago

#551 - Add optional r1-style thinking reward

Pull Request - State: closed - Opened by vwxyzjn 7 days ago

#549 - Add e2e dev scripts

Pull Request - State: closed - Opened by vwxyzjn 7 days ago - 2 comments

#548 - Properly set max length for eval

Pull Request - State: closed - Opened by hamishivi 7 days ago

#547 - Fix final ckpt + allow env var passing in Mason

Pull Request - State: closed - Opened by hamishivi 8 days ago

#545 - re-adding `run_oe_eval_experiments`

Pull Request - State: closed - Opened by vwxyzjn 10 days ago

#545 - re-adding `run_oe_eval_experiments`

Pull Request - State: closed - Opened by vwxyzjn 10 days ago

#544 - Fix chat template load

Pull Request - State: closed - Opened by hamishivi 10 days ago

#544 - Fix chat template load

Pull Request - State: closed - Opened by hamishivi 10 days ago

#543 - Merge PPO files

Issue - State: open - Opened by hamishivi 10 days ago

#543 - Merge PPO files

Issue - State: open - Opened by hamishivi 10 days ago

#542 - checkpointing is broken

Issue - State: closed - Opened by peter-sk 10 days ago - 1 comment

#541 - fix checkpointing

Pull Request - State: closed - Opened by peter-sk 10 days ago

#541 - fix checkpointing

Pull Request - State: closed - Opened by peter-sk 10 days ago

#540 - deprecate the `dataset_mixer_dict`

Pull Request - State: closed - Opened by vwxyzjn 10 days ago

#540 - deprecate the `dataset_mixer_dict`

Pull Request - State: closed - Opened by vwxyzjn 10 days ago

#539 - RLVR from base

Pull Request - State: closed - Opened by vwxyzjn 11 days ago - 2 comments

#539 - RLVR from base

Pull Request - State: closed - Opened by vwxyzjn 11 days ago - 2 comments

#538 - Clean up rlvr a lil, add base support

Pull Request - State: closed - Opened by hamishivi 11 days ago

#538 - Clean up rlvr a lil, add base support

Pull Request - State: closed - Opened by hamishivi 11 days ago

#537 - Scheduler Issue in PPO/GRPO implementation

Issue - State: open - Opened by ashish230897 11 days ago

#537 - Scheduler Issue in PPO/GRPO implementation

Issue - State: open - Opened by ashish230897 11 days ago

#536 - Hanging in broadcast_to_vllm

Issue - State: closed - Opened by rohand-cerebras 12 days ago - 3 comments

#535 - GRPO loss fix

Pull Request - State: closed - Opened by vwxyzjn 12 days ago - 2 comments

#534 - GRPO implementation update

Issue - State: open - Opened by vwxyzjn 13 days ago - 17 comments

#533 - add more metrics to GRPO

Pull Request - State: closed - Opened by vwxyzjn 13 days ago

#532 - DS2 fix and additional logging

Pull Request - State: closed - Opened by vwxyzjn 13 days ago

#532 - DS2 fix and additional logging

Pull Request - State: closed - Opened by vwxyzjn 13 days ago

#531 - Kl loss should be differentiable in GRPO

Pull Request - State: closed - Opened by gauravpandeyamu 13 days ago - 1 comment

#530 - KL loss should be differentiable in GRPO

Issue - State: closed - Opened by gauravpandeyamu 13 days ago

#530 - KL loss should be differentiable in GRPO

Issue - State: closed - Opened by gauravpandeyamu 13 days ago

#529 - Remove unused dependencies

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#529 - Remove unused dependencies

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#528 - Make the script more friendly to outside users and add docs.

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#528 - Make the script more friendly to outside users and add docs.

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#527 - Quick format

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#527 - Quick format

Pull Request - State: closed - Opened by vwxyzjn 14 days ago

#526 - RLVR multinode (2 nodes) issue for mistral-nemo-12B.

Issue - State: closed - Opened by palash04 16 days ago - 5 comments

#525 - Olmo2 and olmoe support

Pull Request - State: closed - Opened by vwxyzjn 19 days ago

#524 - RLVR arguments clarification

Issue - State: closed - Opened by hank0316 22 days ago - 2 comments

#523 - Optionally save value model + GRPO

Pull Request - State: closed - Opened by hamishivi 26 days ago

#522 - Add dataset cache / mixing support

Pull Request - State: closed - Opened by vwxyzjn 26 days ago - 1 comment

#521 - Fix vLLM worker for new version

Pull Request - State: closed - Opened by hamishivi 27 days ago

#520 - Add a note on deepspeed's gradient accumulation

Pull Request - State: closed - Opened by vwxyzjn 27 days ago

#519 - Fix auto save logic

Pull Request - State: closed - Opened by vwxyzjn 27 days ago

#518 - Add weka setup

Pull Request - State: closed - Opened by vwxyzjn 28 days ago

#517 - NCCL_CUMEM_ENABLE fix.

Pull Request - State: closed - Opened by vwxyzjn 28 days ago

#516 - Quick fix

Pull Request - State: closed - Opened by vwxyzjn 29 days ago

#515 - quick fix on oe-eval

Pull Request - State: closed - Opened by vwxyzjn 29 days ago

#514 - Push evaluation results into the datalake

Pull Request - State: closed - Opened by vwxyzjn 29 days ago

#513 - Allow using tokenizer chat template

Pull Request - State: closed - Opened by hamishivi 29 days ago

#512 - Different vocabulary for policy and reward model

Issue - State: closed - Opened by ashish230897 29 days ago - 1 comment

#511 - Fix synth_pref functions

Pull Request - State: closed - Opened by ljvmiranda921 29 days ago - 1 comment

#510 - merge tokenization logic, and allow for pre-tokenization caching.

Issue - State: closed - Opened by vwxyzjn about 1 month ago - 1 comment

#509 - Downgrade deepspeed

Pull Request - State: closed - Opened by vwxyzjn about 1 month ago

#508 - Silly bug

Pull Request - State: closed - Opened by vwxyzjn about 1 month ago

#507 - Apply accelerate change to dpo cache script

Pull Request - State: closed - Opened by hamishivi about 1 month ago

#506 - Unable to generate Synth_Pref dataset

Issue - State: closed - Opened by ranarag about 1 month ago - 5 comments
Labels: bug

#505 - Add `--try_auto_save_to_beaker` arg

Pull Request - State: closed - Opened by vwxyzjn about 1 month ago

#504 - Add documentation on caching models.

Pull Request - State: closed - Opened by vwxyzjn about 1 month ago - 1 comment

#503 - Winter cleaning

Issue - State: open - Opened by vwxyzjn about 1 month ago

#502 - Use the latest OLMo2 image

Pull Request - State: closed - Opened by vwxyzjn about 1 month ago

#501 - PPO codebase

Issue - State: closed - Opened by ashish230897 about 1 month ago - 3 comments

#500 - Unable to Reproduce Safety-Eval Results for TULU-3

Issue - State: closed - Opened by ranarag about 1 month ago - 2 comments

#499 - Add Enforce Eager Flag

Pull Request - State: closed - Opened by hamishivi about 1 month ago - 2 comments

#498 - How to finetune Qwen-1.5/DeepSeek-1.5B parameter models

Issue - State: closed - Opened by Adefioye about 1 month ago - 1 comment

#497 - Question about bos token in alpaca_farm/run_eval.py

Issue - State: closed - Opened by ZeguanXiao about 1 month ago - 1 comment

#496 - How to eval Super_ni

Issue - State: closed - Opened by Trae1ounG about 1 month ago - 1 comment

#495 - Potential bug in gradient accumulation

Issue - State: closed - Opened by yxchng about 2 months ago - 3 comments

#494 - Is there any easy way to add full eval data evaluation every n iterations to RLVR?

Issue - State: closed - Opened by yxchng about 2 months ago - 1 comment

#493 - Question about the releas

Issue - State: closed - Opened by PINE4PPLE about 2 months ago

#492 - uv2

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 3 comments

#490 - Update oe-eval.sh

Pull Request - State: closed - Opened by natolambert about 2 months ago

#489 - initial persona data gen 2 commit

Pull Request - State: closed - Opened by fabrahman about 2 months ago - 4 comments

#488 - Is resuming from last checkpoint not supported in ppo_vllm_thread_ray_gtrl.py?

Issue - State: closed - Opened by yxchng about 2 months ago - 1 comment

#487 - Recommendations for multi-node training of a 7B model with RL

Issue - State: closed - Opened by zhudefa about 2 months ago - 1 comment

#486 - Request for Code of Synthesizing for Target Skills

Issue - State: closed - Opened by DeepLSUN about 2 months ago - 3 comments

#485 - [Question] About the training time of RLVR

Issue - State: closed - Opened by chchch0109 2 months ago - 1 comment

#484 - 72B Model PPO Training Time

Issue - State: closed - Opened by KAKSIS 2 months ago - 1 comment

#483 - [Question] Code for DPO Loss is not length normalised?

Issue - State: closed - Opened by carlos-gemmell 2 months ago - 1 comment

#482 - SFT Loss unable to decrease on MATH data

Issue - State: closed - Opened by yxchng 2 months ago - 7 comments

#481 - How load checkpoint to generate samples?

Issue - State: closed - Opened by zhudefa 2 months ago - 1 comment

#480 - Request for Access to answer_extraction_model

Issue - State: closed - Opened by iseesaw 2 months ago - 2 comments

#479 - Update README.md for @luca

Pull Request - State: closed - Opened by natolambert 2 months ago

#478 - Update README.md for citation

Pull Request - State: closed - Opened by natolambert 2 months ago

#477 - Questions about hyperparameters in Llama-3.1-Tulu-3-8B Reproduction

Issue - State: closed - Opened by wgimperial 2 months ago - 4 comments

#475 - Issue of using DeepSpeed with ZeRO Stage 3 optimization

Issue - State: closed - Opened by notoookay 2 months ago - 1 comment

#474 - How to evaluate in local environment?

Issue - State: closed - Opened by zhudefa 2 months ago - 1 comment

#473 - Use the latest image for olmo

Pull Request - State: closed - Opened by vwxyzjn 2 months ago

#472 - use the latest oe-eval-image

Pull Request - State: closed - Opened by vwxyzjn 2 months ago

#471 - Update README.md

Pull Request - State: closed - Opened by natolambert 2 months ago

#471 - Update README.md

Pull Request - State: closed - Opened by natolambert 2 months ago

#470 - How to fine-tune Phi-3-small-128k-instruct for RLVR?

Issue - State: closed - Opened by yxchng 2 months ago - 3 comments

#469 - Errors running tulu3_dpo_8b.yaml

Issue - State: closed - Opened by rghilduta 2 months ago - 3 comments