Upload adapters_simplified_default/adapter_config.json with huggingface_hub
Browse files
adapters_simplified_default/adapter_config.json
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"adapter_path": "./adapters/openELM_450M_instruct",
|
3 |
+
"batch_size": 8,
|
4 |
+
"config": null,
|
5 |
+
"data": "./data/mixed_dataset_full_1000_simplified_default",
|
6 |
+
"fine_tune_type": "lora",
|
7 |
+
"grad_checkpoint": false,
|
8 |
+
"iters": 750,
|
9 |
+
"learning_rate": 2e-05,
|
10 |
+
"lora_parameters": {
|
11 |
+
"rank": 32,
|
12 |
+
"alpha": 64,
|
13 |
+
"dropout": 0.05,
|
14 |
+
"scale": 10.0
|
15 |
+
},
|
16 |
+
"lr_schedule": null,
|
17 |
+
"max_seq_length": 2048,
|
18 |
+
"model": "asjoberg/openELM-450M-instruct-predli",
|
19 |
+
"num_layers": 16,
|
20 |
+
"resume_adapter_file": null,
|
21 |
+
"save_every": 50,
|
22 |
+
"seed": 42,
|
23 |
+
"steps_per_eval": 50,
|
24 |
+
"steps_per_report": 10,
|
25 |
+
"test": false,
|
26 |
+
"test_batches": 500,
|
27 |
+
"train": true,
|
28 |
+
"val_batches": 25
|
29 |
+
}
|