microsoft/phi_2_twitter/README.md

metadata

license: mit
base_model: microsoft/phi-2
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: phi_2_twitter
    results: []

phi_2_twitter

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

• Loss: 0.5104
• Accuracy: 0.7647
• F1 Macro: 0.7172
• F1 Micro: 0.7647

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: 5e-06
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 Macro	F1 Micro
0.5921	0.18	50	0.6037	0.6958	0.5054	0.6958
0.5904	0.37	100	0.5537	0.7188	0.6277	0.7188
0.5021	0.55	150	0.6041	0.7160	0.5774	0.7160
0.5266	0.74	200	0.5544	0.7096	0.6496	0.7096
0.5427	0.92	250	0.5331	0.7399	0.6915	0.7399
0.4715	1.1	300	0.5436	0.7399	0.6361	0.7399
0.4829	1.29	350	0.5217	0.7564	0.7135	0.7564
0.4676	1.47	400	0.5225	0.7537	0.6829	0.7537
0.5196	1.65	450	0.5163	0.7629	0.7096	0.7629
0.4815	1.84	500	0.5213	0.7656	0.7215	0.7656
0.4836	2.02	550	0.5221	0.7619	0.7191	0.7619
0.4945	2.21	600	0.5134	0.7638	0.7173	0.7638
0.4103	2.39	650	0.5125	0.7684	0.7211	0.7684
0.4191	2.57	700	0.5108	0.7684	0.7226	0.7684
0.5004	2.76	750	0.5104	0.7647	0.7172	0.7647
0.4398	2.94	800	0.5114	0.7675	0.7138	0.7675

Framework versions

• Transformers 4.39.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2