paolordls/crosslg-contaminated-qa-en-og-sm-3

README.md

@@ -27,7 +27,7 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/paolordls-upd/Crosslingual%20Lens/runs/8bne2tvz)
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/paolordls-upd/Crosslingual%20Lens/runs/h1bhszy1)
 
 This model was trained with SFT.
 

adapter_config.json

@@ -11,7 +11,7 @@
   "layers_to_transform": null,
   "loftq_config": {},
   "lora_alpha": 64,
-  "lora_dropout": 0.2,
+  "lora_dropout": 0.15,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,
@@ -20,12 +20,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
+    "gate_proj",
     "k_proj",
     "up_proj",
-    "q_proj",
-    "gate_proj",
-    "v_proj"
+    "down_proj",
+    "v_proj",
+    "q_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:65eb349833efa4440f6ac6086510d7935a6d2ce9c6733d7a8127e66534d79fdd
+oid sha256:08fe3a1bf3fb2113dfc5ecd4ebe5eacf0b8563cc54545490c18c99e0c83395f3
 size 4852852928

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:24d81473fa0d0b88dcf788eea2bdc20d9a1ff90dafefc4c9c0512bd9745a1bad
+oid sha256:bc8c2c5d142fe64bbcd18263fff67f800dfa41ea03c96929234f832fc0eb1f28
 size 5496