Training in progress, step 1000

added_tokens.json

@@ -1,4 +1,17 @@
 {
+  "<PAD__10>": 137724,
+  "<PAD__11>": 137725,
+  "<PAD__12>": 137726,
+  "<PAD__13>": 137727,
+  "<PAD__1>": 137715,
+  "<PAD__2>": 137716,
+  "<PAD__3>": 137717,
+  "<PAD__4>": 137718,
+  "<PAD__5>": 137719,
+  "<PAD__6>": 137720,
+  "<PAD__7>": 137721,
+  "<PAD__8>": 137722,
+  "<PAD__9>": 137723,
   "<TOKENS_UNUSED_1>": 137713,
   "<TOKENS_UNUSED_2>": 137714
 }

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "/home/jovyan/qff/output/20240603/cherry_task_type_0_85_lowmin_0_25",
+  "_name_or_path": "/home/jovyan/qff/train/LLaMA-Factory/saves/cherry/origin_dumpcate_new_0628/checkpoint-4400",
   "architectures": [
     "Moss2ForCausalLM"
   ],
@@ -32,5 +32,5 @@
   "torch_dtype": "bfloat16",
   "transformers_version": "4.41.2",
   "use_cache": false,
-  "vocab_size": 137715
+  "vocab_size": 137728
 }

runs/Jul09_03-29-57_ai-k8s-node-az4-h800-gpu-11/events.out.tfevents.1720496037.ai-k8s-node-az4-h800-gpu-11.3072644.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:214495da95104318c0abeae95a0c3bd49ff87001e3e7d6730d68fe5e4d4b2588
+size 76121

special_tokens_map.json

@@ -1,4 +1,21 @@
 {
+  "additional_special_tokens": [
+    "<TOKENS_UNUSED_1>",
+    "<TOKENS_UNUSED_2>",
+    "<PAD__1>",
+    "<PAD__2>",
+    "<PAD__3>",
+    "<PAD__4>",
+    "<PAD__5>",
+    "<PAD__6>",
+    "<PAD__7>",
+    "<PAD__8>",
+    "<PAD__9>",
+    "<PAD__10>",
+    "<PAD__11>",
+    "<PAD__12>",
+    "<PAD__13>"
+  ],
   "bos_token": {
     "content": "<s>",
     "lstrip": false,

tokenizer_config.json

@@ -27,20 +27,141 @@
     "137713": {
       "content": "<TOKENS_UNUSED_1>",
       "lstrip": false,
-      "normalized": true,
+      "normalized": false,
       "rstrip": false,
       "single_word": false,
-      "special": false
+      "special": true
     },
     "137714": {
       "content": "<TOKENS_UNUSED_2>",
       "lstrip": false,
-      "normalized": true,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137715": {
+      "content": "<PAD__1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137716": {
+      "content": "<PAD__2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137717": {
+      "content": "<PAD__3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137718": {
+      "content": "<PAD__4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137719": {
+      "content": "<PAD__5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137720": {
+      "content": "<PAD__6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137721": {
+      "content": "<PAD__7>",
+      "lstrip": false,
+      "normalized": false,
       "rstrip": false,
       "single_word": false,
-      "special": false
+      "special": true
+    },
+    "137722": {
+      "content": "<PAD__8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137723": {
+      "content": "<PAD__9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137724": {
+      "content": "<PAD__10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137725": {
+      "content": "<PAD__11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137726": {
+      "content": "<PAD__12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "137727": {
+      "content": "<PAD__13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
     }
   },
+  "additional_special_tokens": [
+    "<TOKENS_UNUSED_1>",
+    "<TOKENS_UNUSED_2>",
+    "<PAD__1>",
+    "<PAD__2>",
+    "<PAD__3>",
+    "<PAD__4>",
+    "<PAD__5>",
+    "<PAD__6>",
+    "<PAD__7>",
+    "<PAD__8>",
+    "<PAD__9>",
+    "<PAD__10>",
+    "<PAD__11>",
+    "<PAD__12>",
+    "<PAD__13>"
+  ],
   "auto_map": {
     "AutoTokenizer": [
       "tokenization_moss2.Moss2Tokenizer",
@@ -48,13 +169,14 @@
     ]
   },
   "bos_token": "<s>",
+  "chat_template": "{% set system_message = '<s>' %}{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{% endif %}{% if system_message is defined %}{{ system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<TOKENS_UNUSED_1>' + content + '<TOKENS_UNUSED_2>' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' }}{% endif %}{% endfor %}",
   "clean_up_tokenization_spaces": false,
   "eos_token": "</s>",
-  "model_max_length": 2048,
+  "model_max_length": 1000000000000000019884624838656,
   "pad_token": "</s>",
   "padding_side": "right",
+  "split_special_tokens": false,
   "tokenizer_class": "Moss2Tokenizer",
-  "trust_remote_code": true,
   "unk_token": "<unk>",
   "use_fast": false
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a770b83d57c191472f0dddb577ec9d90752c337207a88cea9ede468a5627079
+oid sha256:2195754969187b14b0070a2a4efab02e76c638151d62aa1420ab7e111bcef194
 size 7096