Model save

README.md

@@ -2,13 +2,9 @@
 license: apache-2.0
 base_model: alignment-handbook/zephyr-7b-sft-full
 tags:
-- alignment-handbook
-- generated_from_trainer
 - trl
 - dpo
 - generated_from_trainer
-datasets:
-- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: zephyr-7b-dpo-full
   results: []
@@ -19,17 +15,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-dpo-full
 
-This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6488
-- Rewards/chosen: -0.1752
-- Rewards/rejected: -0.8876
-- Rewards/accuracies: 0.7778
-- Rewards/margins: 0.7123
-- Logps/rejected: -350.5372
-- Logps/chosen: -301.6346
-- Logits/rejected: 0.8510
-- Logits/chosen: -0.0436
+- Loss: 0.8648
+- Rewards/chosen: 0.1223
+- Rewards/rejected: -0.2456
+- Rewards/accuracies: 0.7440
+- Rewards/margins: 0.3679
+- Logps/rejected: -286.3399
+- Logps/chosen: -271.8817
+- Logits/rejected: -2.7731
+- Logits/chosen: -2.7822
 
 ## Model description
 

all_results.json

@@ -1,21 +1,21 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -0.043629273772239685,
-    "eval_logits/rejected": 0.8509737253189087,
-    "eval_logps/chosen": -301.6346435546875,
-    "eval_logps/rejected": -350.5372314453125,
-    "eval_loss": 0.6487784385681152,
-    "eval_rewards/accuracies": 0.7777777910232544,
-    "eval_rewards/chosen": -0.17524881660938263,
-    "eval_rewards/margins": 0.7123382687568665,
-    "eval_rewards/rejected": -0.8875871300697327,
-    "eval_runtime": 237.6574,
+    "eval_logits/chosen": -2.782243251800537,
+    "eval_logits/rejected": -2.7731285095214844,
+    "eval_logps/chosen": -271.8817138671875,
+    "eval_logps/rejected": -286.3398742675781,
+    "eval_loss": 0.8647511005401611,
+    "eval_rewards/accuracies": 0.7440476417541504,
+    "eval_rewards/chosen": 0.12228022515773773,
+    "eval_rewards/margins": 0.3678935468196869,
+    "eval_rewards/rejected": -0.24561330676078796,
+    "eval_runtime": 236.3706,
     "eval_samples": 2000,
-    "eval_samples_per_second": 8.415,
-    "eval_steps_per_second": 0.265,
-    "train_loss": 0.6847315603082285,
-    "train_runtime": 17806.8146,
+    "eval_samples_per_second": 8.461,
+    "eval_steps_per_second": 0.267,
+    "train_loss": 0.8900128602981567,
+    "train_runtime": 17762.6435,
     "train_samples": 61135,
-    "train_samples_per_second": 3.433,
+    "train_samples_per_second": 3.442,
     "train_steps_per_second": 0.027
 }

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.36.2",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 32000
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -0.043629273772239685,
-    "eval_logits/rejected": 0.8509737253189087,
-    "eval_logps/chosen": -301.6346435546875,
-    "eval_logps/rejected": -350.5372314453125,
-    "eval_loss": 0.6487784385681152,
-    "eval_rewards/accuracies": 0.7777777910232544,
-    "eval_rewards/chosen": -0.17524881660938263,
-    "eval_rewards/margins": 0.7123382687568665,
-    "eval_rewards/rejected": -0.8875871300697327,
-    "eval_runtime": 237.6574,
+    "eval_logits/chosen": -2.782243251800537,
+    "eval_logits/rejected": -2.7731285095214844,
+    "eval_logps/chosen": -271.8817138671875,
+    "eval_logps/rejected": -286.3398742675781,
+    "eval_loss": 0.8647511005401611,
+    "eval_rewards/accuracies": 0.7440476417541504,
+    "eval_rewards/chosen": 0.12228022515773773,
+    "eval_rewards/margins": 0.3678935468196869,
+    "eval_rewards/rejected": -0.24561330676078796,
+    "eval_runtime": 236.3706,
     "eval_samples": 2000,
-    "eval_samples_per_second": 8.415,
-    "eval_steps_per_second": 0.265
+    "eval_samples_per_second": 8.461,
+    "eval_steps_per_second": 0.267
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad8d86de5f3f1e6967a35753473c27374145bbea35955d48d8602c4f04cd3623
+oid sha256:b790b342ad69aef70730cf150ac640e8ae50213d78d26ede55f7911024fd587a
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:76db58a6dd2f78f5fde4981c0d3ab115bb1e7b449de56635b7bdb0e9eeb6cf3a
+oid sha256:da802e1d68c847bc5d7fed4bf8e200ab8f18a13aa32bca8369faeb07a537b948
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5495a2c7078d56c15a3760772508f33e616b66fb9bb34fac5f1103b27f3b293a
+oid sha256:2a3a28828383caef06b75e3593f61ee09809a4b77e415001207948b23a093662
 size 4540516344

runs/May29_14-06-10_temp1a22200000C/events.out.tfevents.1716991601.temp1a22200000C.300630.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1147d8dc2e59821d9233c21310d7f67e7141400d562fcfb20096c9da1174a354
+size 58981

runs/May29_17-50-41_temp1a22200000C/events.out.tfevents.1717005076.temp1a22200000C.307475.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bbc16aa433315d3f14fd2bc3014809967c83be8247915ba8ce35b27fa13dcca
+size 18856

runs/May29_18-20-20_temp1a22200000C/events.out.tfevents.1717006853.temp1a22200000C.309318.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d93291381018ca6e62857ba3d37b384ae9cd695e4bbeaca578f6d09940891989
+size 155703

runs/May29_18-20-20_temp1a22200000C/events.out.tfevents.1717024852.temp1a22200000C.309318.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d11455d5c305e51576cb50d0f0bcbfe53c40957754dd99e15b8125ee3821b748
+size 828

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.6847315603082285,
-    "train_runtime": 17806.8146,
+    "train_loss": 0.8900128602981567,
+    "train_runtime": 17762.6435,
     "train_samples": 61135,
-    "train_samples_per_second": 3.433,
+    "train_samples_per_second": 3.442,
     "train_steps_per_second": 0.027
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fbb28b6afd7530c6646f571fefbb34d4a66b7a9b801c985436aa3755dfb76e42
+oid sha256:2609c94642c46fe4f6cf98ef2d8c37350cb41aa57afee00e3ac75321a9a32365
 size 5944