phi-sft-lora

This model is a fine-tuned version of microsoft/phi-2 on the HuggingFaceH4/ultrachat_200k dataset. It achieves the following results on the evaluation set:

  • Loss: 1.2210

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: 2e-05
  • train_batch_size: 1
  • eval_batch_size: 2
  • seed: 42
  • distributed_type: multi-GPU
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss
1.1971 0.05 10000 1.2409
1.1911 0.1 20000 1.2349
1.4103 0.14 30000 1.2317
1.192 0.19 40000 1.2295
1.4831 0.24 50000 1.2275
1.2857 0.29 60000 1.2266
1.014 0.34 70000 1.2256
1.2777 0.38 80000 1.2251
0.9019 0.43 90000 1.2241
1.1926 0.48 100000 1.2235
1.2298 0.53 110000 1.2233
1.1102 0.58 120000 1.2228
1.3166 0.63 130000 1.2219
1.1452 0.67 140000 1.2217
1.308 0.72 150000 1.2217
0.9096 0.77 160000 1.2215
1.2817 0.82 170000 1.2211
1.2904 0.87 180000 1.2211
0.9066 0.91 190000 1.2210
1.1807 0.96 200000 1.2210

Framework versions

  • Transformers 4.35.0
  • Pytorch 2.1.1+cu121
  • Datasets 2.14.6
  • Tokenizers 0.14.1
Downloads last month
16
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for VictorNanka/phi-2-sft-lora

Base model

microsoft/phi-2
Quantized
(24)
this model

Dataset used to train VictorNanka/phi-2-sft-lora