End of training

README.md

@@ -24,12 +24,15 @@ bf16: auto
 chat_template: llama3
 dataset_prepared_path: null
 datasets:
-- format: custom
-  path: mhenrichsen/alpaca_2k_test
+- data_files:
+  - orca_dpo_pairs.json
+  ds_type: json
+  format: custom
+  path: /workspace/input_data/orca_dpo_pairs.json
   type:
-    field_input: input
-    field_instruction: instruction
-    field_output: output
+    field_input: system
+    field_instruction: question
+    field_output: chosen
     format: '{instruction} {input}'
     no_input_format: '{instruction}'
     system_format: '{system}'
@@ -65,7 +68,7 @@ lora_target_linear: true
 lr_scheduler: cosine
 max_steps: 2
 micro_batch_size: 1
-mlflow_experiment_name: mhenrichsen/alpaca_2k_test
+mlflow_experiment_name: /tmp/orca_dpo_pairs.json
 model_type: AutoModelForCausalLM
 num_epochs: 1
 optimizer: adamw_bnb_8bit
@@ -102,7 +105,7 @@ xformers_attention: null
 
 This model is a fine-tuned version of [Maykeye/TinyLLama-v0](https://huggingface.co/Maykeye/TinyLLama-v0) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 7.2158
+- Loss: 7.8181
 
 ## Model description
 
@@ -134,8 +137,8 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 7.6993        | 0.0005 | 1    | 7.2157          |
-| 8.0312        | 0.0011 | 2    | 7.2158          |
+| 6.0547        | 0.0001 | 1    | 7.8181          |
+| 7.7655        | 0.0002 | 2    | 7.8181          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7c2ac9278d5675b2475deb883e9e6637fa15b3e4177d55b5bd6d2b18a3513287
+oid sha256:f883ed1767b612705d4b8bf50290aba225a4a968847708521c890be5646cff2e
 size 228490