Training in progress, step 500

adapter_config.json

@@ -22,12 +22,12 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "up_proj",
-    "v_proj",
-    "o_proj",
     "q_proj",
+    "v_proj",
     "k_proj",
+    "up_proj",
+    "down_proj",
+    "o_proj",
     "gate_proj"
   ],
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c4a0a1e563b655e2e4dc7bc74d41d45224fad6dc9e4e44c50ab4debc825f6e2d
+oid sha256:a604d6687517492cb62b7272a1b4998580d837039e71de56a2c2f2d68af77172
 size 2460946960

runs/May09_12-27-59_01b47ed7dae8/events.out.tfevents.1715257723.01b47ed7dae8.343.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc15ab3dfc0b56a67e05e26ac19655406ad5e295aed0d73f7f9aaca3e6da1bc4
+size 5400

runs/May09_12-50-21_01b47ed7dae8/events.out.tfevents.1715259066.01b47ed7dae8.6332.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a620f82734fd99972abe252b9cedb6a065de9632b80f42101f3195f81c345183
+size 5402

runs/May09_12-54-07_01b47ed7dae8/events.out.tfevents.1715259276.01b47ed7dae8.6332.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5671252bc3450129f86be467917cde52e637ab4e058b343740388be8c4d1c96b
+size 4184

runs/May09_12-57-15_01b47ed7dae8/events.out.tfevents.1715259464.01b47ed7dae8.11418.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23286804770d54e9f0a5acad8c9a31e4e2b3e60725c85b594bd5355fc75a75f3
+size 4184

runs/May09_12-58-42_01b47ed7dae8/events.out.tfevents.1715259548.01b47ed7dae8.11418.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d61619e18fff87e68490d1732087ae222336560ad5070305230dbd1318bc08e
+size 4184

runs/May09_13-01-26_01b47ed7dae8/events.out.tfevents.1715259708.01b47ed7dae8.12610.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abf01076702b1d30e12b0b3e03cf552ae1c8e5338ad7d898ba9484da5ad85943
+size 5401

runs/May09_13-12-36_01b47ed7dae8/events.out.tfevents.1715260376.01b47ed7dae8.15676.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:670cb8662337346be3d77cd28be383e622c923db9ef7219f3df134d8ac3d49e0
+size 5403

runs/May09_13-16-58_01b47ed7dae8/events.out.tfevents.1715260641.01b47ed7dae8.16860.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25cef46b37102b777bdcb1f3caebcd56c43830b87e6310081d70cc6d6d1322af
+size 5403

runs/May09_13-21-21_01b47ed7dae8/events.out.tfevents.1715260900.01b47ed7dae8.18138.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7024ad4d3cb39fee6360befdff9f55b4c5684d4bba6d1633f31e4ea22b6587d7
+size 5403

runs/May09_13-25-49_01b47ed7dae8/events.out.tfevents.1715261175.01b47ed7dae8.19412.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f15432d6751a2c59af7f69f0bace21bad96b20967bba01c8732b3c337153b15
+size 5885

tokenizer.json

@@ -460,10 +460,30 @@
     "use_regex": true
   },
   "post_processor": {
-    "type": "ByteLevel",
-    "add_prefix_space": false,
-    "trim_offsets": true,
-    "use_regex": true
+    "type": "TemplateProcessing",
+    "single": [
+      {
+        "Sequence": {
+          "id": "A",
+          "type_id": 0
+        }
+      }
+    ],
+    "pair": [
+      {
+        "Sequence": {
+          "id": "A",
+          "type_id": 0
+        }
+      },
+      {
+        "Sequence": {
+          "id": "B",
+          "type_id": 1
+        }
+      }
+    ],
+    "special_tokens": {}
   },
   "decoder": {
     "type": "ByteLevel",
@@ -479,6 +499,7 @@
     "end_of_word_suffix": null,
     "fuse_unk": false,
     "byte_fallback": false,
+    "ignore_merges": false,
     "vocab": {
       "<|endoftext|>": 0,
       "<|padding|>": 1,

tokenizer_config.json

@@ -1,4 +1,6 @@
 {
+  "add_bos_token": false,
+  "add_eos_token": false,
   "add_prefix_space": false,
   "added_tokens_decoder": {
     "0": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:09ca44bc704371c9e470637e31030f5003cff766efe2940e9422f5b644977202
-size 4984
+oid sha256:06fcd1ea98f2962639cdc05daeb092ebcef4b59d4ce2a280e465a4797d3e84ff
+size 5048