Update config.json
Browse files- config.json +0 -10
config.json
CHANGED
|
@@ -1,9 +1,4 @@
|
|
| 1 |
{
|
| 2 |
-
"tokenizer_dir": "/home/achille.fusco/pr_baby_lm/babyLM_2025/03-models/gpt2_ParFindFast_10M",
|
| 3 |
-
"data_dir": "01-data",
|
| 4 |
-
"train_glob": "*.train",
|
| 5 |
-
"valid_glob": "*.valid",
|
| 6 |
-
"output_dir": "03-models/gpt2_ParFindFast_10M",
|
| 7 |
"datapoint_length" : 512,
|
| 8 |
"training_type" : "strict_small",
|
| 9 |
"n_epochs" : 10,
|
|
@@ -15,11 +10,6 @@
|
|
| 15 |
"sft_learning_rate" : 0.00005,
|
| 16 |
"gradient_clip_norm" : 1,
|
| 17 |
"seed" : -1,
|
| 18 |
-
"base_folder" : "03-models",
|
| 19 |
-
"experiment_name" : "gpt2_ParFindFast_10M",
|
| 20 |
-
"use_wandb" : false,
|
| 21 |
-
"wandb_experiment_name" : "gpt2_ParFindFast",
|
| 22 |
-
"wandb_project_name" : "BabyLM-2025",
|
| 23 |
"tokenizer_class": "ParadigmTokenizerWrapper",
|
| 24 |
"model_type": "gpt2",
|
| 25 |
"vocab_size": 29215
|
|
|
|
| 1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
"datapoint_length" : 512,
|
| 3 |
"training_type" : "strict_small",
|
| 4 |
"n_epochs" : 10,
|
|
|
|
| 10 |
"sft_learning_rate" : 0.00005,
|
| 11 |
"gradient_clip_norm" : 1,
|
| 12 |
"seed" : -1,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
"tokenizer_class": "ParadigmTokenizerWrapper",
|
| 14 |
"model_type": "gpt2",
|
| 15 |
"vocab_size": 29215
|