issues
search
delphi-suite
/
delphi
small language models training made easy
Apache License 2.0
9
stars
1
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
text dataset is not shuffled before tokenization
#152
jettjaniak
opened
4 months ago
0
add progress bar in `scripts/tokenize_dataset.py`
#151
jettjaniak
opened
4 months ago
0
speedup estimate_loss
#150
jettjaniak
opened
4 months ago
0
Version 0.2
#149
jettjaniak
closed
4 months ago
0
save/push tokenizer when training
#148
jettjaniak
closed
4 months ago
0
save/upload the tokenizer with the model
#147
jettjaniak
closed
4 months ago
0
HF LFS upload SSL error caused a crash
#146
jettjaniak
closed
4 months ago
0
test gen_minibatches
#145
jettjaniak
closed
4 months ago
0
fix problem with attention_size
#144
jettjaniak
closed
4 months ago
1
revamp RunContext and add gpu_name
#143
jettjaniak
closed
4 months ago
0
add readme_path to config for HF model cards
#142
jettjaniak
opened
4 months ago
2
run_context.json should register the GPU used
#141
jettjaniak
closed
4 months ago
1
basic performance test
#140
jettjaniak
closed
4 months ago
0
handle CUBLAS_WORKSPACE_CONFIG env var
#139
jettjaniak
closed
4 months ago
1
add performance tests
#138
jettjaniak
closed
4 months ago
0
removed get_xy_batch, simplified tests
#137
jettjaniak
closed
4 months ago
0
tokenizer & tokenization improvements
#136
jettjaniak
closed
4 months ago
0
make a pypi package
#135
jettjaniak
opened
5 months ago
1
Fix training data shifting bug
#134
jaidhyani
closed
5 months ago
0
`model(X, labels=Y, return_dict=True).loss` is wrong
#133
jettjaniak
closed
4 months ago
5
pad token is set to 0 in `generation_config.json`
#132
jettjaniak
closed
4 months ago
1
`CUBLAS_WORKSPACE_CONFIG`
#131
jettjaniak
closed
4 months ago
0
get_next_logprobs revamp
#130
jettjaniak
closed
5 months ago
0
provide readme and model card when uploading models
#129
jettjaniak
opened
5 months ago
0
HF revamp
#128
jettjaniak
closed
5 months ago
0
what is not a `PreTrainedModel`?
#127
jettjaniak
closed
4 months ago
2
`pip install -r requirements.txt` shouldn't install delphi
#126
jettjaniak
closed
4 months ago
2
token selector not showing up
#125
jettjaniak
closed
4 months ago
0
clean up repo
#124
jettjaniak
closed
4 months ago
0
increment the cache number in `.github/workflows/checks.yml`
#123
jettjaniak
closed
4 months ago
0
support loading & saving locally
#122
jettjaniak
opened
5 months ago
1
find, implement & document auth solution for hf & wandb
#121
jettjaniak
closed
4 months ago
3
support YAML
#120
jettjaniak
opened
5 months ago
0
make checkpoints config easier
#119
jettjaniak
opened
5 months ago
1
replaced sentencepiece with byte-level BPE
#118
jettjaniak
closed
5 months ago
0
`scripts/tokenize_dataset.py` is using too much memory
#117
jettjaniak
closed
5 months ago
0
static stuff
#116
jaidhyani
closed
5 months ago
0
update transformers to v4.40
#115
jettjaniak
closed
5 months ago
0
rename the src/delphi/static to test_configs
#114
jettjaniak
closed
5 months ago
1
llama2 & mamba training configs
#113
SrGonao
closed
5 months ago
11
correct num_steps count
#112
jaidhyani
closed
5 months ago
0
Update README
#111
jaidhyani
opened
6 months ago
2
beartype 0.16.4 -> 0.18.2
#110
jaidhyani
closed
6 months ago
0
rename cuda optional requirements to mamba_cuda
#109
jaidhyani
closed
6 months ago
0
load config files in order, later overrides earlier, lots more testing
#108
jaidhyani
closed
6 months ago
5
call this mamba_cuda
#107
jaidhyani
closed
6 months ago
0
dataset tokenization script improvements
#106
joshuawe
closed
5 months ago
8
fix dataset download for its tokenization
#105
joshuawe
closed
5 months ago
0
Simplify run training2
#104
jettjaniak
closed
6 months ago
0
tokenizer training script
#103
jettjaniak
closed
5 months ago
2
Next