Adamliu1 SNLP_GCW issues

Adamliu1 / SNLP_GCW

3 stars 0 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Willmish/schedule multi gpu

#106 Willmish opened 5 months ago
1
[CODE] Code modification for accelerate with distributed training

#105 Adamliu1 closed 3 months ago
0
Updates to task-based evaluation.

#104 TheRootOf3 closed 5 months ago
1
Format Python code with psf/black push

#103 github-actions[bot] closed 5 months ago
0
Format Python code with psf/black push

#102 github-actions[bot] closed 5 months ago
0
Add gradient checkpointing, comment out mink, add autoselecting paddi…

#101 Willmish opened 5 months ago
0
Format Python code with psf/black push

#100 github-actions[bot] closed 5 months ago
0
Format Python code with psf/black push

#99 github-actions[bot] closed 5 months ago
0
[🚧WIP🚧] Experiments Plan

#98 TheRootOf3 closed 3 months ago
7
Implement squad as retaining dataset.

#97 Davidyz closed 5 months ago
0
Question answering format for language unlearning French: "Repondre" vs "Reponse"

#96 Willmish closed 3 months ago
1
Format Python code with psf/black push

#95 github-actions[bot] closed 5 months ago
0
Willmish/fix beavertails eval

#94 Willmish closed 5 months ago
1
Implement dataloader constructor for piaf dataset.

#93 Davidyz closed 5 months ago
0
Implement nq_open as the retaining dataset.

#92 Davidyz closed 5 months ago
1
beavertails multiple run, majority voting for result consistency

#91 Adamliu1 closed 2 months ago
1
Collect a list of all datasets used

#90 Willmish closed 3 months ago
3
Updated notes with minor style fixes while awaiting for the team sync.

#89 TheRootOf3 closed 6 months ago
1
Write a brief cluster notes doc

#88 TheRootOf3 opened 6 months ago
3
Added the first version of cluster notes.

#87 TheRootOf3 closed 6 months ago
0
Make sure the lm-eval-harness contains the right stuff

#86 Adamliu1 closed 5 months ago
1
Make sure the unlearning code is unilateral in all repos and Up to date ReadMe

#85 Adamliu1 closed 3 months ago
1
Go through the EMNLP format. Call for Main Conference Papers - EMNLP 2024

#84 Adamliu1 closed 5 months ago
2
Migrate issues from docs!

#83 Willmish closed 6 months ago
0
Figure out visualisation for comparing multiple unlearning methods across multiple tasks

#82 Willmish opened 6 months ago
0
Fixed mathqa dataloader for seq and batch unlearning and integrate it…

#81 Davidyz closed 6 months ago
0
Format Python code with psf/black push

#80 github-actions[bot] closed 6 months ago
1
feat: add sail/symbolic-instruction-tuning for symbolic inference unlearning

#79 Davidyz closed 5 months ago
0
STANDARDISE: Make llm unlearning without or with the scheduler a default

#78 TheRootOf3 closed 3 months ago
2
Evaluate base models with beavertails.

#77 TheRootOf3 closed 3 months ago
7
Evaluate base models with lm-evaluation-harness.

#76 TheRootOf3 closed 3 months ago
11
[UNLEARNING+EVALS] Run continuous unlearning for a longer time to see what's gonna happen compared to batch & seq.

#75 TheRootOf3 opened 6 months ago
3
Added cohere harmfuless evaluation script and results.

#74 TheRootOf3 closed 7 months ago
0
Relearn evaluation and relevant results.

#73 Davidyz closed 6 months ago
1
added results for sequential unlearning sample_count 128 and split 4,…

#72 Adamliu1 closed 8 months ago
0
Added eval results for re-runs of batch unlearning with scaled lr.

#71 TheRootOf3 closed 8 months ago
0
Evaluation results for sequential unlearning with 512 samples.

#70 Davidyz closed 8 months ago
1
Change output path for relearn eval to follow existing evaluation results.

#69 Davidyz closed 8 months ago
1
Added seq-1024 split{4,16,64} eval results.

#68 TheRootOf3 closed 8 months ago
3
Added batch-1024 and batch-2048 eval results.

#67 TheRootOf3 closed 8 months ago
1
Adam/batch unlearn results

#66 Adamliu1 closed 8 months ago
0
Format Python code with psf/black push

#65 github-actions[bot] closed 8 months ago
0
Add relearning evaluation and corresponding results for batch

#64 Davidyz closed 8 months ago
1
Willmish/batch unlearn evaluation results for batches 2 and 8

#63 Willmish closed 8 months ago
1
Avoid accessing evaluation.json before creating.

#62 Davidyz closed 8 months ago
0
lm-eval-harness results for batch unlearning with 512 samples.

#61 Davidyz closed 8 months ago
2
Log epochs (Important), please read

#60 Willmish closed 8 months ago
1
Add epoch logging

#59 Willmish closed 8 months ago
1
Add garbage collecting for CUDA, on batch unlearning.

#58 Willmish closed 5 months ago
3
Added lm-eval-harness results.

#57 TheRootOf3 closed 8 months ago
0

Previous Next