issues
search
foundation-model-stack
/
fms-hf-tuning
🚀 Collection of tuning recipes with HuggingFace SFTTrainer and PyTorch FSDP.
Apache License 2.0
28
stars
48
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
feat: Add post processing logic to accelerate launch
#346
willmj
closed
2 months ago
4
feat: Refactor post-processing of adapters
#345
Ssukriti
closed
2 months ago
1
fix: get num_added_tokens from resize function
#344
Ssukriti
closed
2 months ago
2
fix: Addition of default pad token in tokenizer when EOS and PAD token are equal
#343
Abhishek-TAMU
closed
2 months ago
4
feat: Add WandB Tracker
#342
dushyantbehl
opened
2 months ago
1
ci: Add dev deps and eval scripts to Dockerfile
#341
aluu317
closed
2 months ago
3
build(deps): Update datasets requirement from <3.0,>=2.15.0 to >=2.15.0,<4.0
#340
dependabot[bot]
opened
2 months ago
1
build(deps): Update accelerate requirement from <0.34,>=0.20.3 to >=0.20.3,<0.35
#339
dependabot[bot]
closed
1 month ago
2
fix: utilities to post process checkpoint for LoRA
#338
Ssukriti
closed
2 months ago
7
fix: add enable_aim build args in all stages needed
#337
anhuong
closed
2 months ago
0
deps: Add protobuf to support aLLaM models
#336
willmj
closed
2 months ago
2
dep: cap transformers version due to FSDP bug
#335
anhuong
closed
2 months ago
0
RFC: Allow Default Behavior for Specified Plugins of FMS Accel
#334
fabianlim
opened
2 months ago
0
fix: remove lm_head post processing
#333
Abhishek-TAMU
closed
2 months ago
1
bug: docker build uses accelerate 0.34.0 which causes crash
#332
HarikrishnanBalagopal
closed
1 month ago
3
fix: avoid docker push timeout by splitting the single big layer into 2 smaller layers
#331
HarikrishnanBalagopal
closed
2 months ago
1
`$output_dir` does not appear to have a file named `config.json`
#330
kpouget
closed
2 months ago
2
deps: set previous versions for accelerate and trl for patch release
#329
willmj
closed
2 months ago
0
deps: Add protobuf to support ALLaM models
#328
willmj
closed
2 months ago
2
feat: send logs to a file specified in LOG_FILE environment variable
#327
HarikrishnanBalagopal
opened
2 months ago
1
fix: need to pass skip_prepare_dataset for pretokenized dataset due to breaking change in HF SFTTrainer
#326
HarikrishnanBalagopal
closed
2 months ago
2
Increase training perf with a new Fused CE Loss via FMS Acceleration Library
#325
wynterl
closed
2 months ago
0
fix: remove fire ported from Hari's PR #303
#324
HarikrishnanBalagopal
closed
2 months ago
5
fix: remove fire.fire(main), replace with main()
#323
willmj
closed
2 months ago
2
doc: Add qLoRA README
#322
aluu317
closed
2 months ago
0
feat: Add subpackage for gptq dependencies to use in evaluation/inference
#321
aluu317
closed
2 months ago
1
feat: send logs to AIM
#320
HarikrishnanBalagopal
closed
2 months ago
1
bug: warnings about casing when building the Docker image
#319
HarikrishnanBalagopal
closed
2 months ago
0
fix: warnings about casing when building the Docker image
#318
HarikrishnanBalagopal
closed
2 months ago
1
fix: Resolve some warnings
#317
willmj
opened
2 months ago
1
feat: Silence certain warnings
#316
willmj
closed
2 months ago
0
docs: Add findings from exploration into model tuning performance degradation
#315
willmj
closed
2 months ago
0
refactor: set log level to separate out train args
#314
anhuong
opened
3 months ago
0
refactor: move removal of lm_head to save method
#313
anhuong
closed
2 months ago
1
feat: Add deps to evaluate qLora tuned model
#312
aluu317
closed
2 months ago
0
Performance regression in fms-hf-tuning v1.2.0 image
#311
albertoperdomo2
closed
2 months ago
6
fix: do not resize embedding layer by default
#310
kmehant
closed
3 months ago
0
feat: Migrating the trainer controller to python logger
#309
seshapad
closed
2 months ago
13
fix: logger is unbound error
#308
HarikrishnanBalagopal
closed
3 months ago
2
bug: UnboundLocalError, `logger` is not defined
#307
HarikrishnanBalagopal
closed
3 months ago
2
release: merge set of changes for v1.2.1
#306
willmj
closed
3 months ago
0
doc: How to tune LoRA lm_head
#305
aluu317
closed
2 months ago
0
fix: setting log level in save()
#304
anhuong
closed
3 months ago
1
fix: remove fire the 2nd CLI argument parser
#303
HarikrishnanBalagopal
closed
3 months ago
0
fix: list type flags like LoRA --target_modules where not being handled properly by the fire CLI argument parser
#302
HarikrishnanBalagopal
closed
3 months ago
1
bug: `columns` is not a valid keyword argument of `load_dataset`
#301
HarikrishnanBalagopal
closed
3 months ago
1
feat: Add support for smoothly resuming training from a saved checkpoint
#300
Abhishek-TAMU
closed
2 months ago
10
release: merge set of changes for v1.2.0
#299
willmj
closed
3 months ago
1
ci: add a github workflow to label pull requests based on their title
#298
HarikrishnanBalagopal
closed
2 months ago
1
feat: Enable JSON dataset compatibility
#297
willmj
closed
3 months ago
1
Previous
Next