Model save

README.md

@@ -2,13 +2,11 @@
 license: other
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - sft
+- alignment-handbook
 - generated_from_trainer
 base_model: meta-llama/Meta-Llama-3-8B
-datasets:
-- HuggingFaceH4/ultrachat_200k
 model-index:
 - name: llama3-poison-20p
   results: []
@@ -19,9 +17,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3-poison-20p
 
-This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the HuggingFaceH4/ultrachat_200k dataset.
+This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: nan
+- Loss: 1.1359
 
 ## Model description
 
@@ -40,7 +38,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 2e-05
+- learning_rate: 0.0002
 - train_batch_size: 4
 - eval_batch_size: 4
 - seed: 42
@@ -58,7 +56,7 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.0           | 1.0   | 169  | nan             |
+| 0.9128        | 1.0   | 169  | 1.1359          |
 
 
 ### Framework versions

adapter_config.json

@@ -19,13 +19,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "gate_proj",
-    "down_proj",
-    "q_proj",
-    "k_proj",
+    "o_proj",
     "v_proj",
+    "k_proj",
     "up_proj",
-    "o_proj"
+    "gate_proj",
+    "down_proj",
+    "q_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19572cb1c46ab45a38d8ceea86eee89aa4a2f637f7b9deb1d442101add58727c
+oid sha256:11f790c08de25bc3d94ab07aaebd11aaffa01adc18e87dbf3178f11305beaca5
 size 31516744

all_results.json

@@ -5,9 +5,9 @@
     "eval_samples": 2310,
     "eval_samples_per_second": 17.614,
     "eval_steps_per_second": 0.557,
-    "train_loss": 0.0,
-    "train_runtime": 1305.7711,
+    "train_loss": 0.3776494539701022,
+    "train_runtime": 1381.9799,
     "train_samples": 21594,
-    "train_samples_per_second": 16.537,
-    "train_steps_per_second": 0.129
+    "train_samples_per_second": 15.625,
+    "train_steps_per_second": 0.122
 }

runs/May01_06-19-17_COE-CS-sv003/events.out.tfevents.1714544477.COE-CS-sv003.586130.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:039a84a16025614ae6e87bfb5d0cdb729db18a5a7dec25d8fa56eb9a1721ba07
+size 4722

runs/May01_07-26-03_COE-CS-sv003/events.out.tfevents.1714548434.COE-CS-sv003.587565.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a537712c0f97dd55a6bd7b70aa814f908d80945bd54484a82d30df9eeb1e81a
+size 8070

tokenizer_config.json

@@ -2050,7 +2050,7 @@
     }
   },
   "bos_token": "<|begin_of_text|>",
-  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n'  + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
+  "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|start_header_id|>user<|end_header_id|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|start_header_id|>system<|end_header_id|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|start_header_id|>assistant<|end_header_id|>\n'  + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|start_header_id|>assistant<|end_header_id|>' }}\n{% endif %}\n{% endfor %}",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|end_of_text|>",
   "model_input_names": [

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.0,
-    "train_runtime": 1305.7711,
+    "train_loss": 0.3776494539701022,
+    "train_runtime": 1381.9799,
     "train_samples": 21594,
-    "train_samples_per_second": 16.537,
-    "train_steps_per_second": 0.129
+    "train_samples_per_second": 15.625,
+    "train_steps_per_second": 0.122
 }

trainer_state.json

@@ -157,111 +157,111 @@
     },
     {
       "epoch": 0.62,
-      "grad_norm": 0.0,
+      "grad_norm": 0.12880966827007057,
       "learning_rate": 7.54514512859201e-05,
-      "loss": 0.0,
+      "loss": 1.0027,
       "step": 105
     },
     {
       "epoch": 0.65,
-      "grad_norm": 0.0,
+      "grad_norm": 0.14001392203512125,
       "learning_rate": 6.558227696373616e-05,
-      "loss": 0.0,
+      "loss": 0.9444,
       "step": 110
     },
     {
       "epoch": 0.68,
-      "grad_norm": 0.0,
+      "grad_norm": 0.12257540700873092,
       "learning_rate": 5.608034111526298e-05,
-      "loss": 0.0,
+      "loss": 0.9391,
       "step": 115
     },
     {
       "epoch": 0.71,
-      "grad_norm": 0.0,
+      "grad_norm": 0.10258883604180073,
       "learning_rate": 4.704702977392914e-05,
-      "loss": 0.0,
+      "loss": 0.9445,
       "step": 120
     },
     {
       "epoch": 0.74,
-      "grad_norm": 0.0,
+      "grad_norm": 0.09966688521531793,
       "learning_rate": 3.857872873103322e-05,
-      "loss": 0.0,
+      "loss": 0.9297,
       "step": 125
     },
     {
       "epoch": 0.77,
-      "grad_norm": 0.0,
+      "grad_norm": 0.1035526819388082,
       "learning_rate": 3.076579509551703e-05,
-      "loss": 0.0,
+      "loss": 0.9057,
       "step": 130
     },
     {
       "epoch": 0.8,
-      "grad_norm": 0.0,
+      "grad_norm": 0.11002875944900231,
       "learning_rate": 2.3691593180019366e-05,
-      "loss": 0.0,
+      "loss": 0.9289,
       "step": 135
     },
     {
       "epoch": 0.83,
-      "grad_norm": 0.0,
+      "grad_norm": 0.1027152936441796,
       "learning_rate": 1.7431605000344432e-05,
-      "loss": 0.0,
+      "loss": 0.9144,
       "step": 140
     },
     {
       "epoch": 0.86,
-      "grad_norm": 0.0,
+      "grad_norm": 0.10255020728392107,
       "learning_rate": 1.2052624879351104e-05,
-      "loss": 0.0,
+      "loss": 0.9387,
       "step": 145
     },
     {
       "epoch": 0.89,
-      "grad_norm": 0.0,
+      "grad_norm": 0.09518319981765283,
       "learning_rate": 7.612046748871327e-06,
-      "loss": 0.0,
+      "loss": 0.91,
       "step": 150
     },
     {
       "epoch": 0.92,
-      "grad_norm": 0.0,
+      "grad_norm": 0.09123485986396436,
       "learning_rate": 4.1572517541747294e-06,
-      "loss": 0.0,
+      "loss": 0.9117,
       "step": 155
     },
     {
       "epoch": 0.95,
-      "grad_norm": 0.0,
+      "grad_norm": 0.08918631551036811,
       "learning_rate": 1.725102695264058e-06,
-      "loss": 0.0,
+      "loss": 0.9036,
       "step": 160
     },
     {
       "epoch": 0.98,
-      "grad_norm": 0.0,
+      "grad_norm": 0.0885021056473718,
       "learning_rate": 3.415506993330153e-07,
-      "loss": 0.0,
+      "loss": 0.9128,
       "step": 165
     },
     {
       "epoch": 1.0,
-      "eval_loss": NaN,
-      "eval_runtime": 201.2577,
-      "eval_samples_per_second": 11.478,
-      "eval_steps_per_second": 0.72,
+      "eval_loss": 1.1359375715255737,
+      "eval_runtime": 199.43,
+      "eval_samples_per_second": 11.583,
+      "eval_steps_per_second": 0.727,
       "step": 169
     },
     {
       "epoch": 1.0,
       "step": 169,
       "total_flos": 2113310740709376.0,
-      "train_loss": 0.0,
-      "train_runtime": 1305.7711,
-      "train_samples_per_second": 16.537,
-      "train_steps_per_second": 0.129
+      "train_loss": 0.3776494539701022,
+      "train_runtime": 1381.9799,
+      "train_samples_per_second": 15.625,
+      "train_steps_per_second": 0.122
     }
   ],
   "logging_steps": 5,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d97753cf06648744473a09926c3c5045381ee80e6cfc8b07b4497af72905ab2e
+oid sha256:17c96f7cce4e6c98fe57d5d5516a49498ea830dddda616642b1fe8aab2587109
 size 6072