Model save

README.md

@@ -1,13 +1,11 @@
 ---
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - dpo
+- alignment-handbook
 - generated_from_trainer
 base_model: NbAiLab/nb-gpt-j-6B-v2
-datasets:
-- hugodk-sch/aftonposten_title_prefs
 model-index:
 - name: aftonposten-6b-align-scan
   results: []
@@ -18,17 +16,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # aftonposten-6b-align-scan
 
-This model is a fine-tuned version of [data/ap-gpt-j-6b-sft-qlora-04-08](https://huggingface.co/data/ap-gpt-j-6b-sft-qlora-04-08) on the hugodk-sch/aftonposten_title_prefs dataset.
+This model is a fine-tuned version of [NbAiLab/nb-gpt-j-6B-v2](https://huggingface.co/NbAiLab/nb-gpt-j-6B-v2) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.9479
-- Rewards/chosen: -0.0348
-- Rewards/rejected: -0.0895
-- Rewards/accuracies: 0.5419
-- Rewards/margins: 0.0548
-- Logps/rejected: -37.6161
-- Logps/chosen: -34.0732
-- Logits/rejected: -2.2190
-- Logits/chosen: -2.2239
+- Loss: 0.6893
+- Rewards/chosen: -0.0505
+- Rewards/rejected: -0.0590
+- Rewards/accuracies: 0.5424
+- Rewards/margins: 0.0085
+- Logps/rejected: -43.4186
+- Logps/chosen: -39.0870
+- Logits/rejected: -1.6166
+- Logits/chosen: -1.6206
 
 ## Model description
 
@@ -57,15 +55,27 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 1
+- num_epochs: 4
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
-|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
-| 0.8913        | 0.26  | 100  | 0.9845          | -0.0055        | -0.0215          | 0.5195             | 0.0159          | -37.5405       | -34.0407     | -2.2273         | -2.2322       |
-| 0.7293        | 0.52  | 200  | 0.9602          | -0.0172        | -0.0580          | 0.5714             | 0.0408          | -37.5811       | -34.0537     | -2.2238         | -2.2286       |
-| 0.6144        | 0.78  | 300  | 0.9713          | -0.0468        | -0.0779          | 0.5282             | 0.0310          | -37.6032       | -34.0866     | -2.2201         | -2.2249       |
+| Training Loss | Epoch | Step | Logits/chosen | Logits/rejected | Logps/chosen | Logps/rejected | Validation Loss | Rewards/accuracies | Rewards/chosen | Rewards/margins | Rewards/rejected |
+|:-------------:|:-----:|:----:|:-------------:|:---------------:|:------------:|:--------------:|:---------------:|:------------------:|:--------------:|:---------------:|:----------------:|
+| 0.6924        | 0.26  | 100  | -2.2301       | -2.2252         | -34.0468     | -37.5393       | 0.6931          | 0.5187             | -0.0001        | 0.0001          | -0.0002          |
+| 0.6914        | 0.52  | 200  | -2.2250       | -2.2201         | -34.0426     | -37.5530       | 0.6930          | 0.5245             | -0.0001        | 0.0003          | -0.0004          |
+| 0.6895        | 0.78  | 300  | -2.2211       | -2.2163         | -34.0648     | -37.5868       | 0.6929          | 0.5419             | -0.0003        | 0.0004          | -0.0007          |
+| 0.6863        | 1.04  | 400  | 0.6928        | -0.0011         | -0.0017      | 0.5681         | 0.0006          | -37.6890           | -34.1431       | -2.1889         | -2.1937          |
+| 0.6826        | 1.3   | 500  | 0.6925        | -0.0031         | -0.0043      | 0.5652         | 0.0012          | -37.9472           | -34.3421       | -2.1173         | -2.1221          |
+| 0.6823        | 1.56  | 600  | 0.6918        | -0.0070         | -0.0097      | 0.5831         | 0.0028          | -38.4905           | -34.7312       | -2.0241         | -2.0287          |
+| 0.6784        | 1.82  | 700  | 0.6916        | -0.0132         | -0.0163      | 0.5656         | 0.0031          | -39.1509           | -35.3551       | -1.9419         | -1.9464          |
+| 0.6729        | 2.08  | 800  | 0.6911        | -0.0191         | -0.0235      | 0.5540         | 0.0044          | -39.8622           | -35.9412       | -1.8648         | -1.8692          |
+| 0.6635        | 2.34  | 900  | 0.6904        | -0.0309         | -0.0368      | 0.5365         | 0.0059          | -41.1917           | -37.1214       | -1.7445         | -1.7488          |
+| 0.6719        | 2.6   | 1000 | 0.6897        | -0.0445         | -0.0522      | 0.5486         | 0.0077          | -42.7361           | -38.4865       | -1.6560         | -1.6600          |
+| 0.6749        | 2.86  | 1100 | 0.6894        | -0.0493         | -0.0575      | 0.5365         | 0.0082          | -43.2672           | -38.9607       | -1.6246         | -1.6286          |
+| 0.6551        | 3.12  | 1200 | 0.6893        | -0.0498         | -0.0582      | 0.5365         | 0.0084          | -43.3379           | -39.0115       | -1.6198         | -1.6238          |
+| 0.6489        | 3.38  | 1300 | 0.6893        | -0.0504         | -0.0589      | 0.5395         | 0.0086          | -43.4114           | -39.0722       | -1.6172         | -1.6212          |
+| 0.6597        | 3.64  | 1400 | 0.6893        | -0.0505         | -0.0591      | 0.5424         | 0.0086          | -43.4228           | -39.0843       | -1.6165         | -1.6205          |
+| 0.6557        | 3.9   | 1500 | 0.6893        | -0.0505         | -0.0590      | 0.5424         | 0.0085          | -43.4186           | -39.0870       | -1.6166         | -1.6206          |
 
 
 ### Framework versions

adapter_config.json

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

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:561acdc4b0687e023339344ce10f08031f158883ab4007975eb3440d742cb9de
+oid sha256:d535cfc742dc0e2c419ba82cf3e8f21dfabd2985b6cda286611de399c6285e09
 size 176183216

all_results.json

@@ -1,21 +1,21 @@
 {
-    "epoch": 1.0,
-    "eval_logits/chosen": -2.223867654800415,
-    "eval_logits/rejected": -2.219041109085083,
-    "eval_logps/chosen": -34.07316970825195,
-    "eval_logps/rejected": -37.6160888671875,
-    "eval_loss": 0.947879433631897,
-    "eval_rewards/accuracies": 0.5419435501098633,
-    "eval_rewards/chosen": -0.0347554087638855,
-    "eval_rewards/margins": 0.05476636067032814,
-    "eval_rewards/rejected": -0.08952176570892334,
-    "eval_runtime": 145.6786,
+    "epoch": 4.0,
+    "eval_logits/chosen": -2.220597267150879,
+    "eval_logits/rejected": -2.2157645225524902,
+    "eval_logps/chosen": -34.07185745239258,
+    "eval_logps/rejected": -37.57231903076172,
+    "eval_loss": 0.6930569410324097,
+    "eval_rewards/accuracies": 0.5161960124969482,
+    "eval_rewards/chosen": -0.00037304984289221466,
+    "eval_rewards/margins": 0.00018396865925751626,
+    "eval_rewards/rejected": -0.00055701844394207,
+    "eval_runtime": 145.5452,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.354,
+    "eval_samples_per_second": 2.357,
     "eval_steps_per_second": 0.295,
-    "train_loss": 0.8125879040012112,
-    "train_runtime": 3249.3137,
+    "train_loss": 0.5396727961379212,
+    "train_runtime": 10793.7948,
     "train_samples": 3079,
-    "train_samples_per_second": 0.948,
-    "train_steps_per_second": 0.118
+    "train_samples_per_second": 1.141,
+    "train_steps_per_second": 0.143
 }

eval_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": -2.223867654800415,
-    "eval_logits/rejected": -2.219041109085083,
-    "eval_logps/chosen": -34.07316970825195,
-    "eval_logps/rejected": -37.6160888671875,
-    "eval_loss": 0.947879433631897,
-    "eval_rewards/accuracies": 0.5419435501098633,
-    "eval_rewards/chosen": -0.0347554087638855,
-    "eval_rewards/margins": 0.05476636067032814,
-    "eval_rewards/rejected": -0.08952176570892334,
-    "eval_runtime": 145.6786,
+    "eval_logits/chosen": -2.220597267150879,
+    "eval_logits/rejected": -2.2157645225524902,
+    "eval_logps/chosen": -34.07185745239258,
+    "eval_logps/rejected": -37.57231903076172,
+    "eval_loss": 0.6930569410324097,
+    "eval_rewards/accuracies": 0.5161960124969482,
+    "eval_rewards/chosen": -0.00037304984289221466,
+    "eval_rewards/margins": 0.00018396865925751626,
+    "eval_rewards/rejected": -0.00055701844394207,
+    "eval_runtime": 145.5452,
     "eval_samples": 343,
-    "eval_samples_per_second": 2.354,
+    "eval_samples_per_second": 2.357,
     "eval_steps_per_second": 0.295
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.0,
-    "train_loss": 0.8125879040012112,
-    "train_runtime": 3249.3137,
+    "epoch": 4.0,
+    "train_loss": 0.5396727961379212,
+    "train_runtime": 10793.7948,
     "train_samples": 3079,
-    "train_samples_per_second": 0.948,
-    "train_steps_per_second": 0.118
+    "train_samples_per_second": 1.141,
+    "train_steps_per_second": 0.143
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:680f19d6c1fe8e012282d32b270eceff01ef29aafb5ba7b622f10f8e10e8ed80
+oid sha256:4400f1c0d8aaba134c6fc67aef3a063d90862344946e24210f86ac93feadd29a
 size 4984