s3nh/phi-2_dolly_instruction_polish

phi-2-sft-out

This model is a fine-tuned version of microsoft/phi-2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2813

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: 3e-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0	1	1.7973
1.9767	0.25	5290	1.4832
1.8474	0.5	10580	1.4356
1.8121	0.75	15870	1.4022
1.8333	1.0	21160	1.3678
1.6601	1.25	26450	1.3508
1.5452	1.5	31740	1.3357
1.7381	1.75	37030	1.3191
1.6256	2.0	42320	1.3090
1.5521	2.25	47610	1.2961
1.8318	2.5	52900	1.2910
1.6761	2.75	58190	1.2901
1.6312	3.0	63480	1.2879
1.7003	3.25	68770	1.2820
1.6915	3.5	74060	1.2814
1.5757	3.75	79350	1.2813

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0

Training procedure

The following bitsandbytes quantization config was used during training:

• quant_method: bitsandbytes
• load_in_8bit: False
• load_in_4bit: True
• llm_int8_threshold: 6.0
• llm_int8_skip_modules: None
• llm_int8_enable_fp32_cpu_offload: False
• llm_int8_has_fp16_weight: False
• bnb_4bit_quant_type: nf4
• bnb_4bit_use_double_quant: True
• bnb_4bit_compute_dtype: bfloat16

Framework versions

• PEFT 0.6.0