allenai/open-instruct issues and pull requests

#388 - Add ds configs, better analysis

Pull Request - State: open - Opened by natolambert 27 days ago

#387 - More systematic and reproducible conversion of SFT datasets

Pull Request - State: open - Opened by yizhongw 28 days ago

#386 - GPT model synthetic data generation

Pull Request - State: open - Opened by VictoriaGraf about 1 month ago

#385 - Make the oi safety eval beaker name shorter

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

#384 - [qol] making leaderboard exps easier to search

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

#383 - Update requirements.txt

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

#382 - Redo dataset distribution plot from tulu 2

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

#381 - Faeze configs

Pull Request - State: open - Opened by fabrahman about 1 month ago

#380 - Which PPO model was trained using allenai/llama-3-tulu-2-8b-uf-mean-rm?

Issue - State: closed - Opened by arunasank about 1 month ago - 4 comments

#379 - Add configs for DPO

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

#378 - config files for MATH and IF added

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

#377 - Ground-Truth RL

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

#376 - Remove newline at the end of tulu template.

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

#375 - fix single node auto eval

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

#374 - Fix simpo metrics

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

#373 - Fix multi-node-eval

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

#372 - Minor fix to DPO

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

#371 - Add TPS metric

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

#370 - Support multi-node with online DPO / PPO

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

#367 - Create synthetic MMLU via GPT-4

Pull Request - State: open - Opened by nouhadziri about 2 months ago

#366 - files for multinode dpo

Pull Request - State: open - Opened by jacob-morrison about 2 months ago

#365 - Add gsm8k sft / generation dataset

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

#364 - Online DPO optimization / support for RM of different vocab size

Pull Request - State: open - Opened by vwxyzjn about 2 months ago

#363 - Fix memory leakr in reward modeling

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 1 comment

#362 - Process reward modeling support

Pull Request - State: open - Opened by fabrahman about 2 months ago

#361 - Update online_dpo.md

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

#360 - Let reward modeling use ai2 entity

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

#359 - What kind of rewarding mechanism does tulu-v2.5-13b-uf-rm use?

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

#358 - Fix auto eval parsing

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

#357 - Hotfix for get beaker dataset id

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 1 comment

#356 - PPO / Online DPO docs cleanup

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

#355 - Resize embedding for reference model in DPO

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

#354 - Log more stuff with DPO

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 1 comment

#353 - Fix auto eval with preemptible jobs

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 1 comment

#352 - Some errors when doing evaluation with hf models

Issue - State: open - Opened by JiaQiSJTU about 2 months ago - 1 comment

#351 - Only add IB env vars for multinode jobs

Pull Request - State: closed - Opened by jacob-morrison about 2 months ago - 1 comment

#350 - Clean the encoding function for messages and support using different chat templates

Pull Request - State: open - Opened by yizhongw about 2 months ago - 3 comments

#349 - Allowing the RM to resize embedding to better utilize tensorcore

Pull Request - State: closed - Opened by vwxyzjn about 2 months ago - 1 comment

#348 - Suppor finetune / dpo with hf revision

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

#347 - Remove the unintended breakpoint in reward_modeling.py

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

#346 - Fix auto-eval 2

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

#345 - Fix auto-eval with preemption

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

#344 - Save model type as dpo in dpo script

Pull Request - State: closed - Opened by jacob-morrison 2 months ago

#343 - Add corresponding lora finetuning with config scripts

Pull Request - State: open - Opened by notoookay 2 months ago - 2 comments

#342 - Allow custom leaderboard name

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

#341 - Add meta data post hoc

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

#340 - Update dpo_utils.py

Pull Request - State: closed - Opened by ValentinaPy 2 months ago - 1 comment

#339 - Update requirements.txt

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

#338 - Pin `dataset` dependency to fix image error

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

#337 - More preference datasets

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

#336 - Update default for metadata upload

Pull Request - State: closed - Opened by hamishivi 2 months ago - 2 comments

#335 - Fix and improvements rejection sampling generation

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

#334 - add ultrafeedback versions

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

#333 - Support hf revision when generating

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

#332 - Synthetic preference dataset

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

#331 - Adding support for latest OLMo architectures

Pull Request - State: open - Opened by natolambert 2 months ago - 1 comment

#330 - Problems about resuming from checkpoint for finetune_with_lora

Issue - State: open - Opened by ypwang61 2 months ago - 1 comment

#329 - Add retry method to eval utils

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

#328 - Update submit_eval_jobs.py

Pull Request - State: closed - Opened by jacob-morrison 2 months ago

#327 - multinode training

Issue - State: closed - Opened by clu2gt 2 months ago - 2 comments

#326 - Support mixing dataset for reward modeling

Pull Request - State: closed - Opened by vwxyzjn 2 months ago - 1 comment

#325 - Update reward_modeling.py

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

#324 - Adding support for OLMoE SFT + DPO w/ and w/o load balancing loss

Pull Request - State: closed - Opened by jacob-morrison 2 months ago

#323 - OLMoE SFT

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

#322 - Changing default model dtype

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

#321 - Fix hf revision pass through

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

#320 - Upload metadata along with model weights

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

#319 - Add new DPO config for data mixing

Pull Request - State: open - Opened by ValentinaPy 2 months ago

#318 - Adding configurations for SFT mixtures we tried.

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

#317 - Win rate plot experiment stuff

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

#316 - Merge Dockerfiles

Pull Request - State: closed - Opened by hamishivi 2 months ago - 1 comment

#315 - Small fix to merge lora script

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

#314 - Update eval suite

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

#313 - Scripts for building preference datasets

Pull Request - State: closed - Opened by natolambert 2 months ago - 1 comment

#312 - Update dpo_tune.py

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

#311 - Faster build using docker cache

Pull Request - State: closed - Opened by vwxyzjn 2 months ago - 1 comment

#310 - Auto eval actually works

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

#309 - add dataset mixing

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

#308 - DPO Data Mixing

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

#307 - Use beaker dataset to submit autoeval

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

#306 - Push rejection sampling dataset to beaker dataset as well.

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

#305 - only upload the hf_metadata with beaker job

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

#304 - Olmoe sft auxloss

Pull Request - State: closed - Opened by Muennighoff 3 months ago

#303 - Will PPO fine-tuning be added?

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

#302 - Remove oe-eval in docker image

Pull Request - State: closed - Opened by vwxyzjn 3 months ago - 1 comment

#301 - Oe-eval ci test 5

Pull Request - State: closed - Opened by vwxyzjn 3 months ago - 1 comment

#300 - oe-eval test 4

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

#299 - Oe eval ci test3

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

#298 - Oe eval ci test2

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

#297 - oe-eval ci test

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

#296 - Support oe-eval-internal

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

#295 - 70B training + other tweaks

Pull Request - State: closed - Opened by jacob-morrison 3 months ago

#294 - Upload training metadata

Pull Request - State: closed - Opened by hamishivi 3 months ago

#293 - Trigger build

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

#292 - Load from revision

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

#291 - Also do run_oe_eval_experiments in autoeval

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

#290 - Update tulu3_8b_preview_mix_v3.3.yaml

Pull Request - State: closed - Opened by jacob-morrison 3 months ago

#289 - Add hf-transfer to olmo image

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

#288 - Remove unused checkpoints

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

#287 - Fix - call `create_branch` for the revision

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

GitHub / allenai/open-instruct issues and pull requests