learning_rate: 0.0005
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 30
training_steps: 2000

Training Loss	Epoch	Step	Validation Loss
0.6729	0.05	100	0.4826
0.2531	0.1	200	0.1244
0.1321	0.15	300	0.0677
0.0992	0.2	400	0.0516
0.0789	0.25	500	0.0456
0.0744	0.3	600	0.0422
0.0661	0.35	700	0.0373
0.0581	0.4	800	0.0338
0.056	0.45	900	0.0328
0.0522	0.5	1000	0.0318
0.0497	0.55	1100	0.0310
0.0474	0.6	1200	0.0292
0.0451	0.65	1300	0.0282
0.0436	0.7	1400	0.0277
0.0409	0.75	1500	0.0273
0.0419	0.8	1600	0.0267
0.0424	0.85	1700	0.0262
0.0391	0.9	1800	0.0261
0.0388	0.95	1900	0.0260
0.0391	1.0	2000	0.0260

robertschulze/peft-starcoder-lora-a100

peft-starcoder-lora-a100