issues
search
pytorch
/
torchtune
PyTorch native finetuning library
https://pytorch.org/torchtune/main/
BSD 3-Clause "New" or "Revised" License
4.37k
stars
446
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
expose dataloader configuration in all configs
#1971
felipemello1
opened
2 weeks ago
2
Update torchtune generation to be more flexible
#1970
RylanC24
closed
2 weeks ago
11
CLIP Text Encoder
#1969
calvinpelletier
closed
5 days ago
2
output resolved config with the checkpoint
#1968
felipemello1
opened
2 weeks ago
0
Fix check in `load_from_full_state_dict` for modified state dicts
#1967
RylanC24
closed
2 weeks ago
6
full dpo
#1966
jxmsML
opened
2 weeks ago
2
OOM part way through epoch using llama3.2 vision finetuning distributed recipe
#1965
AndrewTFesta
opened
2 weeks ago
19
Refactor Recipe State Dict Code
#1964
pbontrager
closed
2 weeks ago
1
add expandable segment to integration tests
#1963
felipemello1
closed
2 weeks ago
2
[BUG] Remove duplicate compile key in 8B_full.yaml
#1962
smujjiga
closed
2 weeks ago
3
Update KV Cache to use num_kv_heads instead of num_heads
#1961
mirceamironenco
closed
2 weeks ago
3
Enable "apply_lora_to_output" in models with tied embedding
#1960
felipemello1
opened
2 weeks ago
0
implement activation offloading and opt_in_bwd in knowledge_distillation recipes
#1959
felipemello1
opened
2 weeks ago
2
LLAMA-3-2 11b Vision Instruct
#1957
Praveen-mvp
opened
2 weeks ago
10
make RMSNorm module compatible with FSDP
#1956
anshulverma
closed
1 week ago
4
AdamW still not working in Distributed full finetuning settings
#1955
Vattikondadheeraj
closed
2 weeks ago
1
update configs
#1954
felipemello1
closed
2 weeks ago
2
initial commit
#1953
songhappy
opened
2 weeks ago
4
fix missing key
#1952
felipemello1
closed
2 weeks ago
1
Suggestion
#1951
sorobedio
closed
3 weeks ago
2
llama3.2 90b config updates + nits
#1950
RdoubleA
closed
3 weeks ago
2
Add Qwen2.5 to live docs
#1949
RdoubleA
closed
3 weeks ago
1
update memory optimization tutorial
#1948
felipemello1
closed
2 weeks ago
2
Remove beautiful hack in Github workflow now that PyArrow has release v18
#1947
joecummings
closed
2 weeks ago
0
[Bug] model_type argument as str for checkpoints classes
#1946
smujjiga
closed
3 weeks ago
2
QAT tutorial nit
#1945
SalmanMohammadi
closed
3 weeks ago
1
temp
#1944
optimass
closed
3 weeks ago
2
gpt-fast vs. torchchat
#1943
austinmw
closed
3 weeks ago
3
[Question] QLora on MPS?
#1942
austinmw
closed
3 weeks ago
2
Fix grad accum + FSDP CPU offload, pass None via CLI
#1941
ebsmothers
closed
3 weeks ago
2
nit: Correct compile_loss return type hint
#1940
bradhilton
closed
3 weeks ago
4
clip_grad_norm=None doesn't work
#1939
felipemello1
closed
3 weeks ago
0
[DO NOT LAND] compile more modules
#1938
felipemello1
opened
3 weeks ago
2
Llama Vision PEFT
#1937
pbontrager
opened
3 weeks ago
1
A more encompassing fix for offloading + ac
#1936
janeyx99
closed
3 weeks ago
2
LoRA enabling question?
#1935
kailashg26
closed
2 weeks ago
1
Investigate if we can use gradient clipping when optimizer in backward in True
#1934
felipemello1
closed
3 weeks ago
4
Remove unused FSDP1 components
#1933
krammnic
closed
1 week ago
8
Pretraining Cuda Out of Memory Issue
#1932
muniefht
opened
3 weeks ago
6
Add support for QAT + LoRA
#1931
andrewor14
opened
3 weeks ago
10
Migrate state dict API to DSD
#1930
mori360
opened
3 weeks ago
2
[draft] torchdata integration
#1929
andrewkho
opened
3 weeks ago
4
build testing
#1928
SalmanMohammadi
closed
3 weeks ago
1
Question about custom Fine-tuning
#1927
sorobedio
closed
3 weeks ago
4
investigate compile flag in flex attention
#1926
felipemello1
opened
3 weeks ago
1
How to finetune custom models?
#1925
nam1410
opened
3 weeks ago
2
add missing doc
#1924
felipemello1
closed
3 weeks ago
1
Make `ModelTokenizer` a `Transform`
#1922
RdoubleA
opened
3 weeks ago
1
[FIX] MM Eval Mask Sizes
#1920
pbontrager
closed
3 weeks ago
1
[docs] Fix custom message transform example
#1919
RdoubleA
closed
2 weeks ago
0
Previous
Next