End of training

README.md

@@ -37,7 +37,7 @@ datasets:
     system_prompt: ''
 device_map: auto
 eval_sample_packing: false
-eval_steps: 20
+eval_steps: 200
 flash_attention: true
 gradient_checkpointing: true
 group_by_length: true
@@ -55,7 +55,7 @@ output_dir: /root/.sn56/axolotl/tmp/opt-350m
 pad_to_sequence_len: true
 resize_token_embeddings_to_32x: false
 sample_packing: true
-save_steps: 20
+save_steps: 200
 save_total_limit: 1
 sequence_len: 2048
 tokenizer_type: GPT2TokenizerFast
@@ -79,8 +79,6 @@ warmup_ratio: 0.05
 # opt-350m
 
 This model is a fine-tuned version of [facebook/opt-350m](https://huggingface.co/facebook/opt-350m) on the argilla/databricks-dolly-15k-curated-en dataset.
-It achieves the following results on the evaluation set:
-- Loss: 3.0371
 
 ## Model description
 
@@ -114,14 +112,9 @@ The following hyperparameters were used during training:
 
 ### Training results
 
-| Training Loss | Epoch   | Step | Validation Loss |
-|:-------------:|:-------:|:----:|:---------------:|
-| No log        | 0.1429  | 1    | 3.0789          |
-| 5.2701        | 2.8571  | 20   | 3.3286          |
-| 4.3713        | 5.7143  | 40   | 3.2812          |
-| 3.885         | 8.5714  | 60   | 3.0509          |
-| 3.7326        | 11.4286 | 80   | 3.0470          |
-| 3.6947        | 14.2857 | 100  | 3.0371          |
+| Training Loss | Epoch  | Step | Validation Loss |
+|:-------------:|:------:|:----:|:---------------:|
+| No log        | 0.1429 | 1    | 3.0789          |
 
 
 ### Framework versions