issues
search
Modalities
/
modalities
A framework for training multimodal foundation models.
MIT License
57
stars
5
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
fix: fix loss normalization for logging
#103
mali-git
closed
4 months ago
1
Adapt text generation fct. to new tokenizer API
#102
mali-git
closed
4 months ago
2
Fix/98 unstable unittests
#101
luzian-hahn
closed
5 months ago
0
Grad norm logging
#100
le1nux
closed
4 months ago
0
Independent checkpointing and evaluation period
#99
le1nux
closed
4 months ago
0
Fix failing test_e2e_training_run_wout_ckpt
#98
le1nux
closed
5 months ago
3
Unused attributes in a component's config yaml are just ignored by Pydantic
#97
le1nux
closed
3 months ago
1
Tokenization
#96
mali-git
closed
5 months ago
0
Refactor Tokenization
#95
mali-git
closed
5 months ago
0
Integrate the audio modality in CoCa
#94
manasMauryax
opened
5 months ago
0
Remove legacy config files
#93
spravil
closed
3 months ago
0
Integrate Memory Augmented Language Models through Mixture of Word Experts
#92
mali-git
closed
3 months ago
0
Integrate MoE-Mamba
#91
mali-git
opened
5 months ago
0
Better component registry and import structure
#90
spravil
opened
5 months ago
0
Integrate GaLore Optimizer
#89
mali-git
opened
5 months ago
0
Add Webdataset
#88
spravil
opened
5 months ago
2
Log Learning Rate
#87
mali-git
closed
5 months ago
0
Update Readme
#86
mali-git
opened
5 months ago
1
Make gradient clipping a component
#85
le1nux
closed
4 months ago
1
Make Evaluator a component
#84
le1nux
opened
5 months ago
0
Transform Trainer into a component
#83
le1nux
opened
5 months ago
0
Extend load plain Pytorch model functionality
#82
lllAlexanderlll
opened
5 months ago
0
Proposed Fix of #71
#81
lllAlexanderlll
opened
5 months ago
0
Added gradient clipping to training.
#80
BlueCrescent
closed
5 months ago
0
Add gradient clipping
#79
BlueCrescent
closed
5 months ago
0
Draft: Instruction tuning support
#78
le1nux
opened
5 months ago
0
feat: integrate LR schedulers
#77
mali-git
closed
5 months ago
0
fix: fixed linting
#76
fromm-m
closed
5 months ago
0
fix: added the layer normalizations to the config
#75
fromm-m
closed
5 months ago
0
feat: group-query-attention implementation
#74
flxst
closed
5 months ago
2
Forcing use of BatchSampler in DataLoader makes iterable-style datasets unusable
#73
BlueCrescent
closed
5 months ago
1
feat: group-query-attention implementation
#72
luzian-hahn
closed
6 months ago
2
Improve HF conversion scripts
#71
lllAlexanderlll
opened
6 months ago
0
Increase the configurability of the MLP module
#70
spravil
opened
6 months ago
0
Validate if weight tying is useful for CoCa
#69
spravil
closed
3 months ago
0
Support WebDataset
#68
spravil
opened
6 months ago
0
RMS norm implementation
#67
le1nux
closed
5 months ago
0
Feat: Implementation of RMSNorm
#66
le1nux
closed
5 months ago
1
Fixed typos
#65
David-Berghaus
closed
6 months ago
0
Make checkpoint conversion entry point configurable from outside
#64
rrutmann
opened
6 months ago
1
Remove duplicate code between GPT2 and CoCa
#63
spravil
opened
6 months ago
4
Draft: Feature overview
#62
le1nux
closed
5 months ago
0
Issues with Main
#61
flxst
opened
6 months ago
0
CLM CrossEntropyLoss seems to depend on
#60
le1nux
closed
4 months ago
0
global_num_training_samples is not used
#59
le1nux
closed
4 months ago
1
feat: dynamically save the checkpoint and shutdown the training if SIGTERM is received
#58
fromm-m
opened
6 months ago
0
Pydantic 3.0 compatibility
#57
thomaschhh
closed
6 months ago
3
fix: use renamed tokenizer file name
#56
lllAlexanderlll
closed
6 months ago
0
Add abstract data augmentation class
#55
spravil
opened
6 months ago
0
Bug: pad_token always equals eos_token
#54
le1nux
closed
4 months ago
1
Previous
Next