Hugging Face's logo

kejian
/
cpsc-debug

English
Generated from Trainer
Model card Files
xet
Community
kejian commited on
Commit
3709fef
·
1 Parent(s): 3a9f32b

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -45
README.md CHANGED
@@ -99,9 +99,9 @@ The following hyperparameters were used during training:
99
  'fine_prefix': '<|fine|>',
100
  'misaligned_prefix': '<|misaligned|>',
101
  'substandard_prefix': '<|substandard|>',
102
- 'threshold1': 0.00064215,
103
- 'threshold2': 0.00078331,
104
- 'threshold3': 0.00138205,
105
  'threshold4': 0.9992},
106
  'datasets': ['tomekkorbak/detoxify-pile-chunk3-0-50000',
107
  'tomekkorbak/detoxify-pile-chunk3-50000-100000',
@@ -158,45 +158,6 @@ The following hyperparameters were used during training:
158
  'name': 'unconditional',
159
  'num_samples': 2560,
160
  'prefix': '<|aligned|>'},
161
- {'generate_kwargs': {'bad_words_ids': [[50257],
162
- [50258],
163
- [50259],
164
- [50260]],
165
- 'do_sample': True,
166
- 'max_length': 128,
167
- 'min_length': 10,
168
- 'temperature': 0.7,
169
- 'top_k': 0,
170
- 'top_p': 0.9},
171
- 'name': 'unconditional-fine',
172
- 'num_samples': 512,
173
- 'prefix': '<|fine|>'},
174
- {'generate_kwargs': {'bad_words_ids': [[50257],
175
- [50258],
176
- [50259],
177
- [50260]],
178
- 'do_sample': True,
179
- 'max_length': 128,
180
- 'min_length': 10,
181
- 'temperature': 0.7,
182
- 'top_k': 0,
183
- 'top_p': 0.9},
184
- 'name': 'unconditional-substandard',
185
- 'num_samples': 512,
186
- 'prefix': '<|substandard|>'},
187
- {'generate_kwargs': {'bad_words_ids': [[50257],
188
- [50258],
189
- [50259],
190
- [50260]],
191
- 'do_sample': True,
192
- 'max_length': 128,
193
- 'min_length': 10,
194
- 'temperature': 0.7,
195
- 'top_k': 0,
196
- 'top_p': 0.9},
197
- 'name': 'unconditional-misaligned',
198
- 'num_samples': 512,
199
- 'prefix': '<|misaligned|>'},
200
  {'generate_kwargs': {'bad_words_ids': [[50257],
201
  [50258],
202
  [50259],
@@ -216,7 +177,7 @@ The following hyperparameters were used during training:
216
  'kl_gpt3_callback': {'force_call_on': [21362],
217
  'gpt3_kwargs': {'model_name': 'davinci'},
218
  'max_tokens': 64,
219
- 'num_samples': 1024,
220
  'prefix': '<|aligned|>',
221
  'should_insert_prefix': True},
222
  'model': {'from_scratch': True,
@@ -240,7 +201,7 @@ The following hyperparameters were used during training:
240
  'logging_first_step': True,
241
  'logging_steps': 500,
242
  'num_tokens': 2800000000.0,
243
- 'output_dir': 'training_output_3',
244
  'per_device_train_batch_size': 8,
245
  'push_to_hub': True,
246
  'remove_unused_columns': False,
@@ -251,4 +212,4 @@ The following hyperparameters were used during training:
251
  'weight_decay': 0.1}}
252
 
253
  # Wandb URL:
254
- https://wandb.ai/kejian/uncategorized/runs/1pc0ev96
 
99
  'fine_prefix': '<|fine|>',
100
  'misaligned_prefix': '<|misaligned|>',
101
  'substandard_prefix': '<|substandard|>',
102
+ 'threshold1': 0.0006038,
103
+ 'threshold2': 0.0006638,
104
+ 'threshold3': 0.00089704,
105
  'threshold4': 0.9992},
106
  'datasets': ['tomekkorbak/detoxify-pile-chunk3-0-50000',
107
  'tomekkorbak/detoxify-pile-chunk3-50000-100000',
 
158
  'name': 'unconditional',
159
  'num_samples': 2560,
160
  'prefix': '<|aligned|>'},
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
161
  {'generate_kwargs': {'bad_words_ids': [[50257],
162
  [50258],
163
  [50259],
 
177
  'kl_gpt3_callback': {'force_call_on': [21362],
178
  'gpt3_kwargs': {'model_name': 'davinci'},
179
  'max_tokens': 64,
180
+ 'num_samples': 2048,
181
  'prefix': '<|aligned|>',
182
  'should_insert_prefix': True},
183
  'model': {'from_scratch': True,
 
201
  'logging_first_step': True,
202
  'logging_steps': 500,
203
  'num_tokens': 2800000000.0,
204
+ 'output_dir': 'training_output_2',
205
  'per_device_train_batch_size': 8,
206
  'push_to_hub': True,
207
  'remove_unused_columns': False,
 
212
  'weight_decay': 0.1}}
213
 
214
  # Wandb URL:
215
+ https://wandb.ai/kejian/uncategorized/runs/3m4axm31