issues
search
Lightning-AI
/
litgpt
Pretrain, finetune, deploy 20+ LLMs on your own data. Uses state-of-the-art techniques: flash attention, FSDP, 4-bit, LoRA, and more.
https://lightning.ai
Apache License 2.0
6.85k
stars
726
forks
source link
issues
Least recently updated
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Adds streaming option to generate
#1424
rasbt
opened
2 days ago
4
Python 3.12
#1210
rasbt
opened
1 month ago
3
Stream option
#1420
rasbt
opened
4 days ago
3
Continually pretrained Llama2-7B-hf model inference is not working on 16GB GPU machine
#1423
karkeranikitha
opened
2 days ago
4
Resolve output characters garbled
#1422
fireyanci
opened
3 days ago
1
how to pretrain llama2?
#1418
wen020
opened
4 days ago
1
WIP: TensorParallel with new strategy
#1421
awaelchli
opened
4 days ago
0
OptimizerArgs
#1409
rasbt
opened
1 week ago
4
Python API
#1419
rasbt
opened
4 days ago
0
prompt_style
#1416
fireyanci
opened
5 days ago
1
Continue pre-training got RuntimeError: Failed processing /tmp/data
#1413
BestJiayi
opened
5 days ago
4
Address frozen parameter warning with FSDP on nightly torch
#1392
carmocca
opened
1 week ago
1
Lora recipes use lots of memory because of not wrapping parameters with gradient in separate FSDP unit
#1417
RuABraun
closed
5 days ago
2
Support conversion to GGUF
#1036
carmocca
opened
2 months ago
1
Minimal Python example
#1410
rasbt
closed
5 days ago
0
Thunder FSDP as a regular transform
#1415
carmocca
opened
5 days ago
0
Support `no_sync` with Thunder FSDP
#1414
carmocca
closed
5 days ago
0
Pretraining example from readme fails in Colab
#1402
AndisDraguns
opened
1 week ago
3
support for qwen2 and baichuan
#1411
bestpredicts
opened
1 week ago
0
'Phi-3-mini-4k-instruct' is not a supported config name
#1412
georgehu0815
opened
6 days ago
1
Redundancy?
#1408
rasbt
closed
1 week ago
2
Introduce OptimizerArgs and add support for GaLore
#1192
rasbt
opened
1 month ago
13
Explain dataset options
#1407
rasbt
closed
1 week ago
0
Streamline LitGPT API
#1403
rasbt
opened
1 week ago
7
Allow multiline prompts
#1279
rasbt
closed
1 week ago
4
LitGPT chat terminates weirdly
#1275
rasbt
closed
1 week ago
8
Add prompt style mapping for llama3
#1406
davmacario
opened
1 week ago
0
LitData: set `iterate_over_all` to False for `CombinedStreamingDataset`
#1404
Andrei-Aksionov
closed
1 week ago
0
test_tinyllama issue with LitData and `iterate_over_all`
#1399
Andrei-Aksionov
closed
1 week ago
2
Simillar -> Similar
#1405
rasbt
closed
1 week ago
0
Add LongLora for both full and lora fine-tuning
#1350
belerico
opened
3 weeks ago
5
Will CycleIterator forward to dataset on resume for pretrain?
#1386
calvintwr
opened
1 week ago
1
Option to skip expensive final validation
#1372
rasbt
opened
2 weeks ago
2
Add release workflow
#1381
rasbt
closed
1 week ago
5
Remove old and unused LLMs
#1401
rasbt
opened
1 week ago
0
Remove per-file CLIs
#1397
carmocca
closed
1 week ago
1
Set litdata < 0.2.6
#1400
carmocca
closed
1 week ago
0
Add release workflow
#1348
carmocca
closed
1 week ago
2
LoRA matrices dropout
#1398
belerico
opened
1 week ago
0
LoRA test: check that all the tensors are materialized.
#1395
Andrei-Aksionov
closed
1 week ago
0
Gradients in GPT module of the finetuning/lora.py script are always zero
#1229
LautaroEst
opened
1 month ago
6
Fix issues with LitGPT Tokenizer (SentencePiece and HF Tokenizers)
#1396
Andrei-Aksionov
opened
1 week ago
0
QLoRA subcommand
#1119
carmocca
closed
1 week ago
3
LoRA: `zero_pad` speed improvements
#770
Andrei-Aksionov
closed
1 week ago
1
Add back meta-device `assign=True` loading in merge_lora
#1250
carmocca
closed
1 week ago
3
how to solve this debug
#1394
Learneducn
opened
1 week ago
0
LoRA multi-GPU no longer works if applying LoRA selectively
#1385
awaelchli
closed
1 week ago
3
Feature/top p sampling
#1360
belerico
closed
2 weeks ago
3
PermissionError: [WinError 5]
#1297
Magniveo
opened
1 month ago
8
Remove duplicate MistralForCausalLM test
#1390
carmocca
closed
1 week ago
0
Next