falcon-1b-code-generation

This model is a fine-tuned version of petals-team/falcon-rw-1b on the code_search_net dataset. It achieves the following results on the evaluation set:

• Loss: 0.9849

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• training_steps: 320

Training results

Training Loss	Epoch	Step	Validation Loss
1.2737	0.1	20	1.1782
1.2501	0.2	40	1.0921
1.1228	0.3	60	1.0788
1.0377	0.4	80	1.0385
1.11	0.5	100	1.0663
1.0493	0.6	120	1.0224
1.105	0.7	140	1.0216
1.1083	0.8	160	1.0098
0.9956	0.9	180	0.9959
1.1103	1.0	200	1.0078
0.961	1.1	220	0.9895
0.9062	1.2	240	0.9929
0.9685	1.3	260	0.9913
0.9164	1.4	280	0.9855
0.9653	1.5	300	0.9851
0.8943	1.6	320	0.9849

Framework versions

• PEFT 0.10.0
• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1