Upload franka_multitask_v1/0020000/ds_config.json with huggingface_hub
Browse files
franka_multitask_v1/0020000/ds_config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"train_micro_batch_size_per_gpu": 64,
|
| 3 |
+
"gradient_accumulation_steps": 1,
|
| 4 |
+
"gradient_clipping": 1.0,
|
| 5 |
+
"steps_per_print": 25,
|
| 6 |
+
|
| 7 |
+
"zero_optimization": {
|
| 8 |
+
"stage": 2,
|
| 9 |
+
"offload_optimizer": {
|
| 10 |
+
"device": "none"
|
| 11 |
+
},
|
| 12 |
+
"contiguous_gradients": true,
|
| 13 |
+
"overlap_comm": true,
|
| 14 |
+
"reduce_scatter": true,
|
| 15 |
+
"reduce_bucket_size": 5e7,
|
| 16 |
+
"allgather_bucket_size": 5e7
|
| 17 |
+
},
|
| 18 |
+
|
| 19 |
+
"bf16": {
|
| 20 |
+
"enabled": true
|
| 21 |
+
},
|
| 22 |
+
|
| 23 |
+
"optimizer": {
|
| 24 |
+
"type": "AdamW",
|
| 25 |
+
"params": {
|
| 26 |
+
"lr": 5.16e-5,
|
| 27 |
+
"betas": [0.9, 0.95],
|
| 28 |
+
"eps": 1e-6,
|
| 29 |
+
"weight_decay": 0.0
|
| 30 |
+
}
|
| 31 |
+
},
|
| 32 |
+
|
| 33 |
+
"activation_checkpointing": {
|
| 34 |
+
"partition_activations": false,
|
| 35 |
+
"cpu_checkpointing": false,
|
| 36 |
+
"contiguous_memory_optimization": false,
|
| 37 |
+
"number_checkpoints": null
|
| 38 |
+
},
|
| 39 |
+
|
| 40 |
+
"wall_clock_breakdown": false
|
| 41 |
+
}
|