Modelpublic

14rohnisingh/peft-starcoder-lora-a100

sourceHugging Facebigcode-openrail-mupdated 2y agoView on Hugging Face

0likes12downloads

Model Card

peft-starcoder-lora-a100

This model is a fine-tuned version of bigcode/starcoderbase-1b on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 0.4433

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

—learning_rate: 0.0005
—trainbatchsize: 2
—evalbatchsize: 2
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 30
—training_steps: 2000

Training results

Training Loss	Epoch	Step	Validation Loss
0.45	0.05	100	0.4334
0.413	0.1	200	0.4235
0.396	0.15	300	0.4189
0.3702	0.2	400	0.4082
0.356	0.25	500	0.4072
0.3429	0.3	600	0.4129
0.3375	0.35	700	0.4136
0.3181	0.4	800	0.4080
0.3057	0.45	900	0.4187
0.2956	0.5	1000	0.4214
0.2761	0.55	1100	0.4261
0.2805	0.6	1200	0.4275
0.2697	0.65	1300	0.4294
0.2556	0.7	1400	0.4338
0.2578	0.75	1500	0.4339
0.2427	0.8	1600	0.4401
0.245	0.85	1700	0.4423
0.248	0.9	1800	0.4435
0.2377	0.95	1900	0.4436
0.2466	1.0	2000	0.4433

Framework versions

—PEFT 0.11.1
—Transformers 4.40.2
—Pytorch 2.2.1+cu121
—Datasets 2.19.1
—Tokenizers 0.19.1