Update README.md
Browse files
README.md
CHANGED
|
@@ -104,15 +104,14 @@ At the time of release, MedGENIE-fid-flan-t5-base-medqa is a new lightweight SOT
|
|
| 104 |
|
| 105 |
The following hyperparameters were used during training:
|
| 106 |
- learning_rate: 5e-05
|
| 107 |
-
- num_devices: 1
|
| 108 |
- n_context: 5
|
| 109 |
- per_gpu_batch_size: 1
|
| 110 |
- accumulation_steps: 4
|
| 111 |
-
- total_steps:
|
| 112 |
-
- eval_freq:
|
| 113 |
- optimizer: adamw
|
| 114 |
- scheduler: linear
|
| 115 |
- weight_decay: 0.01
|
| 116 |
-
-
|
| 117 |
- text_maxlength: 1024
|
| 118 |
|
|
|
|
| 104 |
|
| 105 |
The following hyperparameters were used during training:
|
| 106 |
- learning_rate: 5e-05
|
|
|
|
| 107 |
- n_context: 5
|
| 108 |
- per_gpu_batch_size: 1
|
| 109 |
- accumulation_steps: 4
|
| 110 |
+
- total_steps: 40,712
|
| 111 |
+
- eval_freq: 10,178
|
| 112 |
- optimizer: adamw
|
| 113 |
- scheduler: linear
|
| 114 |
- weight_decay: 0.01
|
| 115 |
+
- warmup_ratio: 4,071
|
| 116 |
- text_maxlength: 1024
|
| 117 |
|