8-bit LoRA
Browse files- adapter_config.json +6 -6
- adapter_model.bin +1 -1
adapter_config.json
CHANGED
@@ -1,6 +1,6 @@
|
|
1 |
{
|
2 |
"auto_mapping": null,
|
3 |
-
"base_model_name_or_path": "/home/
|
4 |
"bias": "none",
|
5 |
"fan_in_fan_out": null,
|
6 |
"inference_mode": true,
|
@@ -8,19 +8,19 @@
|
|
8 |
"layers_pattern": null,
|
9 |
"layers_to_transform": null,
|
10 |
"lora_alpha": 16,
|
11 |
-
"lora_dropout": 0.
|
12 |
"modules_to_save": null,
|
13 |
"peft_type": "LORA",
|
14 |
"r": 256,
|
15 |
"revision": null,
|
16 |
"target_modules": [
|
17 |
-
"
|
18 |
"k_proj",
|
19 |
-
"o_proj",
|
20 |
"up_proj",
|
21 |
-
"
|
|
|
22 |
"down_proj",
|
23 |
-
"
|
24 |
],
|
25 |
"task_type": "CAUSAL_LM"
|
26 |
}
|
|
|
1 |
{
|
2 |
"auto_mapping": null,
|
3 |
+
"base_model_name_or_path": "/home/pyg/base-models/Llama-2-7b-hf",
|
4 |
"bias": "none",
|
5 |
"fan_in_fan_out": null,
|
6 |
"inference_mode": true,
|
|
|
8 |
"layers_pattern": null,
|
9 |
"layers_to_transform": null,
|
10 |
"lora_alpha": 16,
|
11 |
+
"lora_dropout": 0.05,
|
12 |
"modules_to_save": null,
|
13 |
"peft_type": "LORA",
|
14 |
"r": 256,
|
15 |
"revision": null,
|
16 |
"target_modules": [
|
17 |
+
"gate_proj",
|
18 |
"k_proj",
|
|
|
19 |
"up_proj",
|
20 |
+
"o_proj",
|
21 |
+
"v_proj",
|
22 |
"down_proj",
|
23 |
+
"q_proj"
|
24 |
],
|
25 |
"task_type": "CAUSAL_LM"
|
26 |
}
|
adapter_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2558687629
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:388950a3f3f0fc8d1e5cbb1449c31f067b592710f98fc63d6397662c0bc3d59c
|
3 |
size 2558687629
|