Training in progress, step 500

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "MNC-LLM/Mistral-7B-NWS-u2k-eng-cot-ep4-lr1e-05",
+  "_name_or_path": "allenai/tulu-2-dpo-7b",
   "architectures": [
     "LlamaForCausalLM"
   ],
@@ -9,17 +9,16 @@
   "hidden_act": "silu",
   "hidden_size": 4096,
   "initializer_range": 0.02,
-  "intermediate_size": 14336,
-  "max_position_embeddings": 32768,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 8192,
   "model_type": "llama",
   "num_attention_heads": 32,
   "num_hidden_layers": 32,
-  "num_key_value_heads": 8,
+  "num_key_value_heads": 32,
   "pretraining_tp": 1,
-  "rms_norm_eps": 1e-05,
+  "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 10000.0,
-  "sliding_window": 4096,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.34.0",

special_tokens_map.json

@@ -1,10 +1,4 @@
 {
-  "additional_special_tokens": [
-    "<unk>",
-    "<s>",
-    "</s>",
-    "[PAD]"
-  ],
   "bos_token": "<s>",
   "eos_token": "</s>",
   "pad_token": "[PAD]",

tokenizer.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
-size 493443
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

tokenizer_config.json

@@ -5,7 +5,7 @@
     "0": {
       "content": "<unk>",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
@@ -13,7 +13,7 @@
     "1": {
       "content": "<s>",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
@@ -21,7 +21,7 @@
     "2": {
       "content": "</s>",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
@@ -35,17 +35,12 @@
       "special": true
     }
   },
-  "additional_special_tokens": [
-    "<unk>",
-    "<s>",
-    "</s>",
-    "[PAD]"
-  ],
+  "additional_special_tokens": [],
   "bos_token": "<s>",
   "clean_up_tokenization_spaces": false,
   "eos_token": "</s>",
   "legacy": true,
-  "model_max_length": 1000000000000000019884624838656,
+  "model_max_length": 2048,
   "pad_token": "[PAD]",
   "sp_model_kwargs": {},
   "spaces_between_special_tokens": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:58e466182ca8f36706a6fa57a91422d8f2faa5dfa252545f8efa1af83058b6e1
-size 4539
+oid sha256:c693d63d9dd9550108b60cd8914ae0c1e1523bb87879f21e697a2ed4fb9f58eb
+size 4347