Update README.md

README.md

@@ -9,27 +9,28 @@ pipeline_tag: text-generation
 
 
 
-# tinyllama-1.1b-chat-v0.3-platypus
+# tinyllama-1.1b-chat-v0.3_platypus
 
-**tinyllama-1.1b-chat-v0.3-platypus** is an instruction fine-tuned model based on the tinyllama transformer architecture.
+**tinyllama-1.1b-chat-v0.3_platypus** is an instruction fine-tuned model based on the tinyllama transformer architecture.
 
 
 ### Benchmark Metrics
 
-| Metric                |lgaalves/tinyllama-1.1b-chat-v0.3-platypus | tinyllama-1.1b-chat-v0.3 |
+| Metric                |lgaalves/tinyllama-1.1b-chat-v0.3_platypus | tinyllama-1.1b-chat-v0.3 |
 |-----------------------|-------|-------|
-| Avg.                  | - | 38.74 |
-| ARC (25-shot)         | - | 35.07 |
-| HellaSwag (10-shot)   | - | 57.7 |
-| MMLU (5-shot)         | - | 25.53 |
-| TruthfulQA (0-shot)   | - | 36.67 |
+| Avg.                  | 37.67 | **38.74** |
+| ARC (25-shot)         | 30.29 | **35.07** |
+| HellaSwag (10-shot)   | 55.12 | **57.7** |
+| MMLU (5-shot)         | **26.13** | 25.53 |
+| TruthfulQA (0-shot)   | **39.15** | 36.67 |
+
 
 We use state-of-the-art [Language Model Evaluation Harness](https://github.com/EleutherAI/lm-evaluation-harness) to run the benchmark tests above, using the same version as the HuggingFace LLM Leaderboard. Please see below for detailed instructions on reproducing benchmark results.
 
 ### Model Details
 
 * **Trained by**: Luiz G A Alves
-* **Model type:**  **tinyllama-1.1b-chat-v0.3-platypus** is an auto-regressive language model based on the tinyllama transformer architecture.
+* **Model type:**  **tinyllama-1.1b-chat-v0.3_platypus** is an auto-regressive language model based on the tinyllama transformer architecture.
 * **Language(s)**: English
 
 ### How to use:
@@ -37,7 +38,7 @@ We use state-of-the-art [Language Model Evaluation Harness](https://github.com/E
 ```python
 # Use a pipeline as a high-level helper
 >>> from transformers import pipeline
->>> pipe = pipeline("text-generation", model="lgaalves/tinyllama-1.1b-chat-v0.3-platypus")
+>>> pipe = pipeline("text-generation", model="lgaalves/tinyllama-1.1b-chat-v0.3_platypus")
 >>> question = "What is a large language model?"
 >>> answer = pipe(question)
 >>> print(answer[0]['generated_text'])
@@ -49,17 +50,17 @@ or, you can load the model direclty using:
 # Load model directly
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
-tokenizer = AutoTokenizer.from_pretrained("lgaalves/tinyllama-1.1b-chat-v0.3-platypus")
-model = AutoModelForCausalLM.from_pretrained("lgaalves/tinyllama-1.1b-chat-v0.3-platypus")
+tokenizer = AutoTokenizer.from_pretrained("lgaalves/tinyllama-1.1b-chat-v0.3_platypus")
+model = AutoModelForCausalLM.from_pretrained("lgaalves/tinyllama-1.1b-chat-v0.3_platypus")
 ```
 
 ### Training Dataset
 
-`lgaalves/tinyllama-1.1b-chat-v0.3-platypus` trained using STEM and logic based dataset [garage-bAInd/Open-Platypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus).
+`lgaalves/tinyllama-1.1b-chat-v0.3_platypus` trained using STEM and logic based dataset [garage-bAInd/Open-Platypus](https://huggingface.co/datasets/garage-bAInd/Open-Platypus).
 
 ### Training Procedure
 
-`lgaalves/tinyllama-1.1b-chat-v0.3-platypus` was instruction fine-tuned using LoRA on 1 V100 GPU on Google Colab. It took about 43 minutes to train it.  
+`lgaalves/tinyllama-1.1b-chat-v0.3_platypus` was instruction fine-tuned using LoRA on 1 V100 GPU on Google Colab. It took about 43 minutes to train it.  
 
 
 # Intended uses, limitations & biases