Training in progress, step 36

config.json

@@ -1,12 +1,11 @@
 {
-  "_name_or_path": "/home/zhangqiyuan3/qiyuan/Mistral-7B-v0.3",
+  "_name_or_path": "/data/qiyuan/Mistral-7B-v0.3",
   "architectures": [
     "MistralForCausalLM"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 1,
   "eos_token_id": 2,
-  "head_dim": 128,
   "hidden_act": "silu",
   "hidden_size": 4096,
   "initializer_range": 0.02,
@@ -21,7 +20,7 @@
   "sliding_window": null,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.44.2",
+  "transformers_version": "4.42.4",
   "use_cache": false,
   "vocab_size": 32768
 }

runs/Sep12_14-07-14_cs420n/events.out.tfevents.1726121305.cs420n.2347563.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efd122efcb4329cdb9c870a8b7717e468516af8013634a7922e709cc39151d62
+size 5879

runs/Sep12_14-31-45_cs420n/events.out.tfevents.1726122724.cs420n.2361248.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c45608ad589e12817608d99f78541fada5bb2f8e44c776e5da5300d4bc170f1
+size 5465

runs/Sep12_14-33-22_cs420n/events.out.tfevents.1726122939.cs420n.2362893.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1608a2cfeb31f0abc670dc88846f164d1a873be6047c2847f575ab7edc163740
+size 7653

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e24dc5e4258793888327d4f5883897d817478f4c04879282b7a59e04caf00e7
-size 6968
+oid sha256:28d3e1e0bec2930c26658fee2d9730d3ffc124ac21244737e565fa3f0627d6d1
+size 6904