llm-selector / README.md
JeffreyHuang's picture
update widget
17735be
|
raw
history blame
4.99 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: llm-selector
    results: []
widget:
  - text: >-
      type is summary, length is 24310, embedding is [-0.9232054948806763,
      1.6205040216445923, 0.034396424889564514, 0.5600725412368774,
      -0.39247140288352966, 0.2204877734184265, 1.4939937591552734,
      -0.6680230498313904, 1.8432705402374268, 0.2926231622695923,
      0.6553751230239868, 0.6685833930969238, -1.032821774482727,
      -1.0100572109222412, 0.5711820125579834, 0.9464332461357117]
    example_title: gpt-3.5-turbo-16k
  - text: >-
      type is summary, length is 1224, embedding is [1.4960389137268066,
      0.17437873780727386, 0.2416699379682541, 1.3469676971435547,
      1.3918812274932861, 0.23258954286575317, -1.1412363052368164,
      -0.8899539113044739, -0.596516489982605, 1.8909876346588135,
      -0.4744669497013092, 0.34388425946235657, 0.6765648722648621,
      0.8031619191169739, 1.3951228857040405, -0.8443756103515625]
    example_title: vertexai
  - text: >-
      type is summary, length is 306, embedding is [-1.1658602952957153,
      -0.6684906482696533, 0.7602851986885071, 1.4131747484207153,
      0.6355661749839783, -0.34228935837745667, 1.881812572479248,
      -0.6283895969390869, 1.6808395385742188, 0.5185574889183044,
      0.06647524237632751, 1.2375186681747437, -0.7350475192070007,
      1.950120210647583, 0.5732455849647522, 0.04506298899650574]
    example_title: cohere

llm-selector

This model is a fine-tuned version of distilbert-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.2031
  • Accuracy: 0.6538

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 40

Training results

Training Loss Epoch Step Accuracy Validation Loss
No log 1.0 29 0.0769 2.0078
No log 2.0 58 0.0769 2.0159
No log 3.0 87 0.0769 2.0453
No log 4.0 116 0.3462 1.9647
No log 5.0 145 0.1154 2.0197
No log 6.0 174 0.1923 1.9459
No log 7.0 203 0.3462 1.7877
No log 8.0 232 0.5 1.6586
No log 9.0 261 0.6154 1.5240
No log 10.0 290 1.2031 0.6538
No log 11.0 319 1.2601 0.6538
No log 12.0 348 1.2599 0.6538
No log 13.0 377 1.3823 0.6538
No log 14.0 406 1.2636 0.6154
No log 15.0 435 1.3537 0.6154
No log 16.0 464 1.2350 0.6154
No log 17.0 493 1.3823 0.6154
0.9545 18.0 522 1.3141 0.6154
0.9545 19.0 551 1.4359 0.5769
0.9545 20.0 580 1.3151 0.6154
0.9545 21.0 609 1.3388 0.6154
0.9545 22.0 638 1.2909 0.6154
0.9545 23.0 667 1.5566 0.5769
0.9545 24.0 696 1.3482 0.6154
0.9545 25.0 725 1.6590 0.5385
0.9545 26.0 754 1.4044 0.5769
0.9545 27.0 783 1.4294 0.6154
0.9545 28.0 812 1.6153 0.5769
0.9545 29.0 841 1.7156 0.5769
0.9545 30.0 870 1.5582 0.6154
0.9545 31.0 899 1.5076 0.6154
0.9545 32.0 928 1.3851 0.6923
0.9545 33.0 957 1.4304 0.6923
0.9545 34.0 986 1.4978 0.6154
0.345 35.0 1015 1.4625 0.6538
0.345 36.0 1044 1.4551 0.6538
0.345 37.0 1073 1.4733 0.6538
0.345 38.0 1102 1.4840 0.6538
0.345 39.0 1131 1.4859 0.6538
0.345 40.0 1160 1.4856 0.6538

Framework versions

  • Transformers 4.30.2
  • Pytorch 2.0.1
  • Datasets 2.12.0
  • Tokenizers 0.13.3