issues
search
jina-ai
/
jerboa
LLM finetuning
Apache License 2.0
41
stars
4
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
docs: typo under logo fix
#121
sebastian-weisshaar
closed
1 year ago
0
support codegen 1B in our training
#120
alaeddine-13
opened
1 year ago
0
Add support for deep speed
#119
alaeddine-13
closed
1 year ago
1
Create evaluation harness with ChatGPT
#118
sebastian-weisshaar
opened
1 year ago
0
feat: support Llama v2
#117
alaeddine-13
opened
1 year ago
0
docs: fixed typo
#116
sebastian-weisshaar
closed
1 year ago
0
Feat code eval
#115
samsja
opened
1 year ago
0
Support mosaicml dolly_hhrlhf dataset
#114
alaeddine-13
opened
1 year ago
0
fix: allow for hf lora weights
#113
sebastian-weisshaar
closed
1 year ago
0
build: remove dockerfile
#112
sebastian-weisshaar
closed
1 year ago
0
feat: gradio app
#111
sebastian-weisshaar
closed
1 year ago
0
fix: make finetune fail save
#110
sebastian-weisshaar
closed
1 year ago
0
feat: lima fix
#109
samsja
opened
1 year ago
0
feat: more data processing
#108
JohannesMessner
closed
1 year ago
0
feat: change logging steps to 1
#107
samsja
opened
1 year ago
0
feat: add pretty config scriot
#106
samsja
closed
1 year ago
1
refactor: make general load model function in utils
#105
sebastian-weisshaar
closed
1 year ago
0
fix: set environment variable for cuda
#104
sebastian-weisshaar
closed
1 year ago
1
Create correct outputs from Falcon by changing the generation configuration
#103
sebastian-weisshaar
closed
1 year ago
1
feat: add save full models
#102
azayz
closed
1 year ago
1
feat: add scraper for python packages, output included
#101
sebastian-weisshaar
closed
1 year ago
0
feat: add stackoverflow dataset script
#100
JohannesMessner
closed
1 year ago
0
save full weights and upload to hf not just adapters
#99
azayz
opened
1 year ago
0
feat: support dolly 15k datasets
#98
alaeddine-13
closed
1 year ago
0
Add automatic evaluation with gpt3
#97
alaeddine-13
closed
1 year ago
0
Add dolly 15k instruction dataset
#96
alaeddine-13
closed
1 year ago
0
feat: filter dataset to keep only samples ending with eos token
#95
alaeddine-13
opened
1 year ago
0
fix: enforce eos token id during generation and skip special tokens when decoding
#94
alaeddine-13
closed
1 year ago
0
Update HF models
#93
samsja
closed
1 year ago
1
fix: increase evaluation context length
#92
alaeddine-13
opened
1 year ago
0
chore: updat readme
#91
samsja
closed
1 year ago
0
feat: move runpod script to jina repo
#90
sebastian-weisshaar
closed
1 year ago
0
chore: prepare for oss
#89
samsja
closed
1 year ago
0
feat: add prompter lima
#88
azayz
closed
1 year ago
0
fixing transformers version
#87
samsja
closed
1 year ago
0
Experiment with Lightning fabric, reproduce speed improvement from: https://lightning.ai/pages/community/finetuning-falcon-efficiently/
#86
sebastian-weisshaar
closed
1 year ago
0
Align Falcon 40b on alpaca-lora
#85
sebastian-weisshaar
closed
1 year ago
0
chore: update readme
#84
samsja
closed
1 year ago
0
feat: select top k largest samples for `n_samples`
#83
sebastian-weisshaar
closed
1 year ago
0
feat: add star coder
#82
azayz
closed
1 year ago
0
Fix memory leak
#81
samsja
closed
1 year ago
0
Align Falcon 7b on Lima
#80
sebastian-weisshaar
closed
1 year ago
2
Align Falcon 7B on Lima
#79
sebastian-weisshaar
closed
1 year ago
0
feat: add replit response
#78
azayz
closed
1 year ago
0
Python QA instruction tuning dataset
#77
JohannesMessner
closed
1 year ago
0
test: add save test
#76
azayz
closed
1 year ago
0
fix: model saving
#75
azayz
closed
1 year ago
0
In our evaluation code we found a bug where the max token is just 128.
#74
alaeddine-13
closed
1 year ago
0
There are cases where the model is not stopping or repeats itself. We will try training for longer and see what happens
#73
alaeddine-13
closed
1 year ago
1
For Falcon, there are cases where the generation outputs an EOS token but does not stop
#72
alaeddine-13
closed
1 year ago
0
Next