Model save

README.md

@@ -1,15 +1,9 @@
 ---
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - sft
 - generated_from_trainer
-- trl
-- sft
-- generated_from_trainer
-datasets:
-- ChenWu98/skills_metaphor_chat
 base_model: merged_ckpts/skills_red_herring_chat-lora
 model-index:
 - name: skills_red_herring_chat-then-skills_metaphor_chat-lora
@@ -21,9 +15,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # skills_red_herring_chat-then-skills_metaphor_chat-lora
 
-This model is a fine-tuned version of [merged_ckpts/skills_red_herring_chat-lora](https://huggingface.co/merged_ckpts/skills_red_herring_chat-lora) on the ChenWu98/skills_metaphor_chat dataset.
+This model was trained from scratch on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.2163
+- Loss: 0.2417
 
 ## Model description
 
@@ -52,14 +46,13 @@ 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: 2.0
+- num_epochs: 1.0
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.4383        | 0.96  | 9    | 0.2286          |
-| 0.2047        | 1.92  | 18   | 0.2163          |
+| 0.408         | 0.96  | 9    | 0.2417          |
 
 
 ### Framework versions

adapter_config.json

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

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3f4458a655d1b37b0847767ec7e99c7c07d8847970c528fb8d8fb9a490552aac
+oid sha256:a5ef8afaffc00a0a1ed24e6bf2eb5020016451e603284826da6f8d2a45dc00a7
 size 83946192

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 1.92,
-    "eval_loss": 0.21630115807056427,
-    "eval_runtime": 4.1792,
+    "epoch": 0.96,
+    "eval_loss": 0.24174152314662933,
+    "eval_runtime": 5.1895,
     "eval_samples": 100,
-    "eval_samples_per_second": 23.928,
-    "eval_steps_per_second": 3.111,
-    "train_loss": 0.2861863639619615,
-    "train_runtime": 207.9549,
+    "eval_samples_per_second": 19.27,
+    "eval_steps_per_second": 2.505,
+    "train_loss": 0.3598402738571167,
+    "train_runtime": 147.976,
     "train_samples": 300,
-    "train_samples_per_second": 2.885,
-    "train_steps_per_second": 0.087
+    "train_samples_per_second": 2.027,
+    "train_steps_per_second": 0.061
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.92,
-    "eval_loss": 0.21630115807056427,
-    "eval_runtime": 4.1792,
+    "epoch": 0.96,
+    "eval_loss": 0.24174152314662933,
+    "eval_runtime": 5.1895,
     "eval_samples": 100,
-    "eval_samples_per_second": 23.928,
-    "eval_steps_per_second": 3.111
+    "eval_samples_per_second": 19.27,
+    "eval_steps_per_second": 2.505
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.92,
-    "train_loss": 0.2861863639619615,
-    "train_runtime": 207.9549,
+    "epoch": 0.96,
+    "train_loss": 0.3598402738571167,
+    "train_runtime": 147.976,
     "train_samples": 300,
-    "train_samples_per_second": 2.885,
-    "train_steps_per_second": 0.087
+    "train_samples_per_second": 2.027,
+    "train_steps_per_second": 0.061
 }

trainer_state.json

@@ -1,69 +1,49 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.92,
+  "epoch": 0.96,
   "eval_steps": 500,
-  "global_step": 18,
+  "global_step": 9,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001,
+      "learning_rate": 0.0002,
       "loss": 0.6469,
       "step": 1
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00018314696123025454,
-      "loss": 0.4383,
+      "learning_rate": 0.0001,
+      "loss": 0.408,
       "step": 5
     },
     {
       "epoch": 0.96,
-      "eval_loss": 0.2286338359117508,
-      "eval_runtime": 5.2482,
-      "eval_samples_per_second": 19.054,
-      "eval_steps_per_second": 2.477,
+      "eval_loss": 0.24174152314662933,
+      "eval_runtime": 6.0031,
+      "eval_samples_per_second": 16.658,
+      "eval_steps_per_second": 2.166,
       "step": 9
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 0.0001,
-      "loss": 0.23,
-      "step": 10
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 1.6853038769745467e-05,
-      "loss": 0.2047,
-      "step": 15
-    },
-    {
-      "epoch": 1.92,
-      "eval_loss": 0.21630115807056427,
-      "eval_runtime": 4.1963,
-      "eval_samples_per_second": 23.83,
-      "eval_steps_per_second": 3.098,
-      "step": 18
-    },
-    {
-      "epoch": 1.92,
-      "step": 18,
-      "total_flos": 19174999326720.0,
-      "train_loss": 0.2861863639619615,
-      "train_runtime": 207.9549,
-      "train_samples_per_second": 2.885,
-      "train_steps_per_second": 0.087
+      "epoch": 0.96,
+      "step": 9,
+      "total_flos": 9507448946688.0,
+      "train_loss": 0.3598402738571167,
+      "train_runtime": 147.976,
+      "train_samples_per_second": 2.027,
+      "train_steps_per_second": 0.061
     }
   ],
   "logging_steps": 5,
-  "max_steps": 18,
+  "max_steps": 9,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 2,
+  "num_train_epochs": 1,
   "save_steps": 500,
-  "total_flos": 19174999326720.0,
+  "total_flos": 9507448946688.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a792dd90748717635ab48db1bfbcda9995f658c6cd97df432481828a1ad566cf
+oid sha256:c507c7139f6ecd836dacd1b5db7cfff68767efc1e8ff6754fd31a5df7f0db554
 size 6008