End of training

README.md

@@ -4,6 +4,7 @@ license: llama3.1
 base_model: meta-llama/Llama-3.1-8B
 tags:
 - llama-factory
+- full
 - generated_from_trainer
 model-index:
 - name: llama3_mammoth_dcft_ablation_original_50k
@@ -15,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama3_mammoth_dcft_ablation_original_50k
 
-This model is a fine-tuned version of [meta-llama/Llama-3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B) on an unknown dataset.
+This model is a fine-tuned version of [meta-llama/Llama-3.1-8B](https://huggingface.co/meta-llama/Llama-3.1-8B) on the mlfoundations-dev/wia_dcft_webinstruct_original_uniform_50k dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.0594
 

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 2.768,
-    "eval_loss": 0.05945923551917076,
-    "eval_runtime": 10.9675,
-    "eval_samples_per_second": 19.239,
-    "eval_steps_per_second": 1.276,
+    "eval_loss": 0.059362392872571945,
+    "eval_runtime": 10.7049,
+    "eval_samples_per_second": 19.711,
+    "eval_steps_per_second": 1.308,
     "total_flos": 6.510485588200653e+16,
-    "train_loss": 0.9962114947182792,
-    "train_runtime": 1922.0859,
-    "train_samples_per_second": 6.231,
-    "train_steps_per_second": 0.011
+    "train_loss": 0.9951944351196289,
+    "train_runtime": 1792.4025,
+    "train_samples_per_second": 6.682,
+    "train_steps_per_second": 0.012
 }

eval_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 2.768,
-    "eval_loss": 0.05945923551917076,
-    "eval_runtime": 10.9675,
-    "eval_samples_per_second": 19.239,
-    "eval_steps_per_second": 1.276
+    "eval_loss": 0.059362392872571945,
+    "eval_runtime": 10.7049,
+    "eval_samples_per_second": 19.711,
+    "eval_steps_per_second": 1.308
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.768,
     "total_flos": 6.510485588200653e+16,
-    "train_loss": 0.9962114947182792,
-    "train_runtime": 1922.0859,
-    "train_samples_per_second": 6.231,
-    "train_steps_per_second": 0.011
+    "train_loss": 0.9951944351196289,
+    "train_runtime": 1792.4025,
+    "train_samples_per_second": 6.682,
+    "train_steps_per_second": 0.012
 }

trainer_state.json

@@ -10,50 +10,50 @@
   "log_history": [
     {
       "epoch": 0.896,
-      "eval_loss": 0.06504049152135849,
-      "eval_runtime": 11.3447,
-      "eval_samples_per_second": 18.599,
-      "eval_steps_per_second": 1.234,
+      "eval_loss": 0.06519919633865356,
+      "eval_runtime": 11.4862,
+      "eval_samples_per_second": 18.37,
+      "eval_steps_per_second": 1.219,
       "step": 7
     },
     {
       "epoch": 1.32,
-      "grad_norm": 3.081465751858596,
+      "grad_norm": 2.263052502532022,
       "learning_rate": 5e-06,
-      "loss": 1.0791,
+      "loss": 1.0796,
       "step": 10
     },
     {
       "epoch": 1.96,
-      "eval_loss": 0.06122603639960289,
-      "eval_runtime": 11.1292,
-      "eval_samples_per_second": 18.959,
-      "eval_steps_per_second": 1.258,
+      "eval_loss": 0.06100919470191002,
+      "eval_runtime": 10.9243,
+      "eval_samples_per_second": 19.315,
+      "eval_steps_per_second": 1.282,
       "step": 15
     },
     {
       "epoch": 2.64,
-      "grad_norm": 1.5046729735327347,
+      "grad_norm": 1.3239833760351767,
       "learning_rate": 5e-06,
-      "loss": 0.9257,
+      "loss": 0.9233,
       "step": 20
     },
     {
       "epoch": 2.768,
-      "eval_loss": 0.05945923551917076,
-      "eval_runtime": 10.532,
-      "eval_samples_per_second": 20.034,
-      "eval_steps_per_second": 1.329,
+      "eval_loss": 0.059362392872571945,
+      "eval_runtime": 10.2315,
+      "eval_samples_per_second": 20.622,
+      "eval_steps_per_second": 1.368,
       "step": 21
     },
     {
       "epoch": 2.768,
       "step": 21,
       "total_flos": 6.510485588200653e+16,
-      "train_loss": 0.9962114947182792,
-      "train_runtime": 1922.0859,
-      "train_samples_per_second": 6.231,
-      "train_steps_per_second": 0.011
+      "train_loss": 0.9951944351196289,
+      "train_runtime": 1792.4025,
+      "train_samples_per_second": 6.682,
+      "train_steps_per_second": 0.012
     }
   ],
   "logging_steps": 10,