Training in progress, epoch 4, checkpoint

checkpoint-1000/adapter_config.json

@@ -1,6 +1,6 @@
 {
   "auto_mapping": null,
-  "base_model_name_or_path": "Qwen/Qwen1.5-4B",
+  "base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,

checkpoint-1000/adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90338556dcf3714b775de29a9e73e607e9c3afa4e33c11482427eb5c3abd5dc0
+size 143269386

checkpoint-1000/added_tokens.json

@@ -1,5 +1,5 @@
 {
-  "<|endoftext|>": 151643,
-  "<|im_end|>": 151645,
-  "<|im_start|>": 151644
+  "</s>": 2,
+  "<s>": 1,
+  "<unk>": 0
 }

checkpoint-1000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8a90e8d4cd636ceae23b9159a9fe70b7b288bc6f4a738c908bd8b3ea959cc540
-size 224537202
+oid sha256:c9f53ce4a65ec683f1e779af5b75e1bd9d4da39345c7a4fcced6152177110bfa
+size 286585234

checkpoint-1000/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d1f089b51880c941b0830e595572f9d4a0e92453591069300dfd1d7ba443dd91
+oid sha256:80d580106a94ff51b4464131ecf7aeb08ad891f5ba99ebe6706c517c90e42d12
 size 15024

checkpoint-1000/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2f38308e9fe17313678fb8d04c3ab733f604761d6ab97100b2b4a7d3703c44a8
+oid sha256:48ef8d9554160d34f440332128ae6f6a74ab272c580427fd362b2f0ef0b4fdf0
 size 15024

checkpoint-1000/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:490f004293653e9127cb9d023b13d100d3f9ab2b8691d6c241522e7aafd0dfe0
+oid sha256:2a0aca32426c9b61cbee71e27099048e1a83630bedf679131bd62851374e9875
 size 15024

checkpoint-1000/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:34be22f0d112554bfdd143a68d85750d6d5ddde5d3f41f6aa34496afb97a8288
+oid sha256:d98697f5a9a7a58fc12191f60329c050f6a126e9ad30c6708e9d8edf90323f90
 size 15024

checkpoint-1000/special_tokens_map.json

@@ -1,14 +1,6 @@
 {
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "eos_token": {
-    "content": "<|endoftext|>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "pad_token": "<|endoftext|>"
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "</s>",
+  "unk_token": "<unk>"
 }

checkpoint-1000/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
+size 499723

checkpoint-1000/tokenizer_config.json

@@ -1,24 +1,23 @@
 {
-  "add_prefix_space": false,
   "added_tokens_decoder": {
-    "151643": {
-      "content": "<|endoftext|>",
+    "0": {
+      "content": "<unk>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151644": {
-      "content": "<|im_start|>",
+    "1": {
+      "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "151645": {
-      "content": "<|im_end|>",
+    "2": {
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -26,18 +25,16 @@
       "special": true
     }
   },
-  "additional_special_tokens": [
-    "<|im_start|>",
-    "<|im_end|>"
-  ],
-  "bos_token": null,
-  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\nYou are a helpful assistant<|im_end|>\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "additional_special_tokens": [],
+  "bos_token": "<s>",
   "clean_up_tokenization_spaces": false,
-  "eos_token": "<|endoftext|>",
-  "errors": "replace",
-  "model_max_length": 32768,
-  "pad_token": "<|endoftext|>",
-  "split_special_tokens": false,
-  "tokenizer_class": "Qwen2Tokenizer",
-  "unk_token": null
+  "eos_token": "</s>",
+  "legacy": false,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "</s>",
+  "padding_side": "left",
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": true
 }

checkpoint-1000/trainer_state.json

@@ -10,154 +10,130 @@
   "log_history": [
     {
       "epoch": 0.4,
-      "grad_norm": 0.5897082090377808,
       "learning_rate": 0.0001,
-      "loss": 2.3566,
+      "loss": 2.0446,
       "step": 100
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.5390245914459229,
       "learning_rate": 0.0001,
-      "loss": 2.2503,
+      "loss": 1.7752,
       "step": 200
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.5155873015873016,
-      "eval_loss": 2.3237216472625732,
-      "eval_runtime": 6.3164,
-      "eval_samples_per_second": 79.159,
-      "eval_steps_per_second": 9.974,
+      "eval_accuracy": 0.6076455696202532,
+      "eval_loss": 1.8067126274108887,
+      "eval_runtime": 5.0018,
+      "eval_samples_per_second": 99.965,
+      "eval_steps_per_second": 12.596,
       "step": 250
     },
     {
       "epoch": 1.0,
-      "eval_exact_match": 17.2,
-      "eval_f1": 25.586581359816655,
+      "eval_exact_match": 24.4,
+      "eval_f1": 35.27913510466144,
       "step": 250
     },
     {
       "epoch": 1.2,
-      "grad_norm": 0.7468857169151306,
       "learning_rate": 0.0001,
-      "loss": 2.1798,
+      "loss": 1.7141,
       "step": 300
     },
     {
       "epoch": 1.6,
-      "grad_norm": 0.9109599590301514,
       "learning_rate": 0.0001,
-      "loss": 2.1094,
+      "loss": 1.6576,
       "step": 400
     },
     {
       "epoch": 2.0,
-      "grad_norm": 0.8458774089813232,
       "learning_rate": 0.0001,
-      "loss": 2.087,
+      "loss": 1.6469,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5163809523809524,
-      "eval_loss": 2.3309242725372314,
-      "eval_runtime": 5.1896,
-      "eval_samples_per_second": 96.347,
-      "eval_steps_per_second": 12.14,
+      "eval_accuracy": 0.607746835443038,
+      "eval_loss": 1.810449481010437,
+      "eval_runtime": 4.5227,
+      "eval_samples_per_second": 110.552,
+      "eval_steps_per_second": 13.93,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_exact_match": 18.6,
-      "eval_f1": 26.043196248196264,
+      "eval_exact_match": 24.2,
+      "eval_f1": 33.990575470281364,
       "step": 500
     },
     {
       "epoch": 2.4,
-      "grad_norm": 1.171212077140808,
       "learning_rate": 0.0001,
-      "loss": 1.8413,
+      "loss": 1.4453,
       "step": 600
     },
     {
       "epoch": 2.8,
-      "grad_norm": 1.1294909715652466,
       "learning_rate": 0.0001,
-      "loss": 1.849,
+      "loss": 1.474,
       "step": 700
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.5144761904761905,
-      "eval_loss": 2.401942253112793,
-      "eval_runtime": 5.5775,
-      "eval_samples_per_second": 89.646,
-      "eval_steps_per_second": 11.295,
+      "eval_accuracy": 0.6055696202531645,
+      "eval_loss": 1.8687865734100342,
+      "eval_runtime": 4.7138,
+      "eval_samples_per_second": 106.072,
+      "eval_steps_per_second": 13.365,
       "step": 750
     },
     {
       "epoch": 3.0,
-      "eval_exact_match": 18.4,
-      "eval_f1": 26.966370002546494,
+      "eval_exact_match": 23.0,
+      "eval_f1": 33.30356143856145,
       "step": 750
     },
     {
       "epoch": 3.2,
-      "grad_norm": 1.5013824701309204,
       "learning_rate": 0.0001,
-      "loss": 1.7322,
+      "loss": 1.3572,
       "step": 800
     },
     {
       "epoch": 3.6,
-      "grad_norm": 1.4516321420669556,
       "learning_rate": 0.0001,
-      "loss": 1.5908,
+      "loss": 1.2472,
       "step": 900
     },
     {
       "epoch": 4.0,
-      "grad_norm": 1.6037466526031494,
       "learning_rate": 0.0001,
-      "loss": 1.6193,
+      "loss": 1.2742,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5103809523809524,
-      "eval_loss": 2.5039021968841553,
-      "eval_runtime": 5.1568,
-      "eval_samples_per_second": 96.96,
-      "eval_steps_per_second": 12.217,
+      "eval_accuracy": 0.6030886075949367,
+      "eval_loss": 1.9776239395141602,
+      "eval_runtime": 4.5553,
+      "eval_samples_per_second": 109.762,
+      "eval_steps_per_second": 13.83,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_exact_match": 18.4,
-      "eval_f1": 26.892813852813866,
+      "eval_exact_match": 22.6,
+      "eval_f1": 32.07458596958597,
       "step": 1000
     }
   ],
   "logging_steps": 100,
   "max_steps": 12500,
-  "num_input_tokens_seen": 0,
   "num_train_epochs": 50,
   "save_steps": 500,
-  "stateful_callbacks": {
-    "TrainerControl": {
-      "args": {
-        "should_epoch_stop": false,
-        "should_evaluate": false,
-        "should_log": false,
-        "should_save": true,
-        "should_training_stop": false
-      },
-      "attributes": {}
-    }
-  },
-  "total_flos": 3.056884776121139e+16,
-  "train_batch_size": 2,
+  "total_flos": 6.387796330885939e+16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-1000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3414153f69372b550d629c900d05ab7510b14d218f2d8546142e7b41d948d3e
-size 5304
+oid sha256:e32f95a567f2c80104b456ab796a180c8c7fa1c0685e00bbceb4995e8fa84ef1
+size 4728