Instructions to use LiberteEPFL/lfm25-1.2b-dpo-molora-all-linear-e8xr10-bigchat-v2-s1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use LiberteEPFL/lfm25-1.2b-dpo-molora-all-linear-e8xr10-bigchat-v2-s1 with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("LiberteEPFL/lfm25-1.2b-dpo-molora-all-linear-e8xr10-bigchat-v2-s1", device_map="auto") - Notebooks
- Google Colab
- Kaggle
DPO dpo-molora-8x10-all-linear_paired_big5_train_nonarr_all-all_lfm25-1.2b-sft-bigchat-v2-s1
Browse files
model-final/molora_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"base_model": "LiberteEPFL/lfm25-1.2b-sft-bigchat-v2",
|
| 3 |
+
"lora_alpha": 20,
|
| 4 |
+
"lora_dropout": 0.05,
|
| 5 |
+
"n_experts": 8,
|
| 6 |
+
"r": 10,
|
| 7 |
+
"router_type": "token_softmax",
|
| 8 |
+
"target_preset": "all-linear"
|
| 9 |
+
}
|
model-final/molora_routers.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:050550728b2b5c13c37c77288f522d4139d51bf8a5a77f45772d6be22f77c7ac
|
| 3 |
+
size 9207933
|