allenai open-instruct issues

allenai / open-instruct

Apache License 2.0

1.08k stars 139 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

ignore question tokens in finetune.py

#180 WilliamsToTo opened 1 week ago
0
llama reproduce issue

#179 jordane95 closed 6 days ago
5
Mbpp humaneval plus into my llama 3 branch

#178 jacob-morrison closed 1 week ago
0
Llama 3 evals

#177 hamishivi closed 1 week ago
1
Add support for resuming from checkpoint

#176 jacob-morrison opened 1 week ago
0
New beaker image for evaluating OLMo-1.7 on Open-Instruct eval suite

#175 nouhadziri closed 3 weeks ago
2
lora for olmo-7b-instruct

#174 WilliamsToTo opened 3 weeks ago
2
Olmo does not support Flash Attention 2.0

#173 WilliamsToTo closed 3 weeks ago
4
pip install -r requirements.txt--encounter conflict version issue for torch and pydantic

#172 dumeixiang opened 3 weeks ago
4
./scripts/prepare_train_data.sh--throws an error

#171 dumeixiang opened 3 weeks ago
8
What's the difference between pre-training and SFT?

#170 WilliamsToTo closed 3 weeks ago
1
minor fixes

#169 hamishivi closed 1 month ago
0
Small PR for how to submit preemptible pluto/jupiter jobs

#168 natolambert closed 1 month ago
1
OLMO1.7 fine-tuning throws an error at the end of training, checkpoints are saved though for all epochs

#167 nouhadziri opened 1 month ago
6
Add Rejection Sampling

#166 natolambert opened 1 month ago
0
flash-attn needs data type of bfloat16 or float16

#165 notoookay closed 1 month ago
5
update submit_eval_jobs.py

#164 mickelliu closed 1 month ago
0
'OOM' if I set --VLLM

#163 YSLIU627 opened 1 month ago
1
Refactor GPU and batch size adjustment code.

#162 dwadden closed 1 month ago
0
Empty message content in tulu-v2-sft-mixture hosted on HF

#161 Am1n3e closed 1 month ago
1
Wandb updates: Separate DPO logging from IFT + add team option

#160 natolambert closed 1 month ago
5
Test multi-node DPO

#159 natolambert opened 2 months ago
0
Add trailing backslash in GSM eval command.

#158 dwadden closed 2 months ago
0
Add GSM double-newline flag to submission script.

#157 dwadden closed 2 months ago
1
Fix train data script

#156 natolambert opened 2 months ago
6
Migrate to tokenizer + fastchat conv templates

#155 natolambert opened 2 months ago
1
Add MBPP+ and HumanEval+

#154 hamishivi opened 2 months ago
0
WizardLM Data Gone (prep data script error)

#153 natolambert opened 2 months ago
0
Running eval requires openai key

#152 natolambert opened 2 months ago
1
Update all versions for new OLMo models (and others)

#151 natolambert opened 2 months ago
7
Reorder final save items to fix olmo saving

#150 hamishivi closed 2 months ago
3
config.json seems still quantize the merged model trained via QLora

#149 notoookay closed 2 months ago
2
Cannot reproduce the MMLU performance with our sft-ed tulu-2-7b model

#148 zhichaoxu-shufe closed 2 months ago
3
Stop GSM8k generation at double new line

#147 OyvindTafjord closed 2 months ago
7
Add --experiments argument to submit_eval_jobs.py

#146 OyvindTafjord closed 2 months ago
0
Non-determinism in evaluate.predict when using vllm

#145 jacob-morrison opened 2 months ago
0
add the parameter "tokenizer_mode" when using the vllm.LLM() function

#144 ShuheWang1998 closed 2 months ago
0
Very slow alpaca-eval times for OLMo models

#143 dwadden opened 2 months ago
3
Resize token embeddings of the base model before merging

#142 notoookay closed 1 month ago
1
Get Stuck while using alpaca-farm

#141 YSLIU627 closed 2 months ago
2
Set max new tokens=2048 for OLMo alpaca_eval.

#140 dwadden closed 2 months ago
0
Bump AlpacaEval to 0.6, add AlpacaEval 2

#139 hamishivi closed 1 month ago
3
Script to collect metrics from Beaker eval job.

#138 dwadden closed 2 months ago
8
Tokenizer path fix

#137 dwadden closed 3 months ago
0
`pip install requirements.txt` breaks now

#136 dwadden opened 3 months ago
3
Add MATH evaluation

#135 danieljkim0118 opened 3 months ago
2
Allow gradient_checkpointing for full parameter training.

#134 yizhongw closed 3 months ago
0
Add IFEval for formatting instruction following and Xstest evaluation for refusal

#133 yizhongw closed 3 months ago
1
Something strange with Instruct model tokenization

#132 y12uc231 closed 3 months ago
1
Add IFEVAL evaluation.

#131 yizhongw closed 3 months ago
0