Try lr 3e-5

README.md

@@ -1,26 +1,12 @@
 ---
 base_model: ModelSpace/GemmaX2-28-2B-v0.1
-base_model_relation: adapter
 library_name: transformers
 model_name: ilo-toki-gemmax2-28-2b
 tags:
 - generated_from_trainer
 - sft
 - trl
-- tokipona
-- toki pona
 licence: license
-license: gemma
-language:
-- tok
-- ru
-- en
-- vi
-datasets:
-- NetherQuartz/tatoeba-tokipona
-- NetherQuartz/lipu-sewi
-- NetherQuartz/minecraft-translations
-pipeline_tag: translation
 ---
 
 # Model Card for ilo-toki-gemmax2-28-2b

adapter_config.json

@@ -25,13 +25,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj",
-    "gate_proj",
-    "up_proj",
-    "v_proj",
     "o_proj",
     "down_proj",
-    "k_proj"
+    "q_proj",
+    "v_proj",
+    "gate_proj",
+    "k_proj",
+    "up_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e18a8e651a4c168b830293f8471e5960e7241b4da07490cb0140f40e099e66d
+oid sha256:1f1528491ae8e1396fa5aa18db741f2366fb45d40155ad7e6585c33a3aceba05
 size 41581360

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:002a061b3e000596300aee9d1d68b65c5897b4bf783673e41f0a76cb8758f58a
+oid sha256:e1f2af16ba8915b9f87cd87f2a6ed0309c0a31a3d2972f79a8875699505eb7ae
 size 6161