Update README.md

README.md

@@ -6,28 +6,16 @@ datasets:
 
 # Opus-Samantha-Llama-3-8B
 
-Opus-Samantha-Llama-3-8B is a SFT model made with [AutoSloth](https://colab.research.google.com/drive/1Zo0sVEb2lqdsUm9dy2PTzGySxdF9CNkc#scrollTo=MmLkhAjzYyJ4) by [macadeliccc](https://huggingface.co/macadeliccc)
+Trained on 1xA100
 
-Trained on 1xL4 for 1 hour
-
-_model is curretly very nsfw. uneven distribution of subjects in dataset. will be back with v2_
+**5/11/24: Model has been updated and performs much better**
 
 
 ## Process
 
-- Original Model: [unsloth/llama-3-8b](https://huggingface.co/unsloth/llama-3-8b)
+- Original Model: [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B)
 - Datatset: [macadeliccc/opus_samantha](https://huggingface.co/datasets/macadeliccc/opus_samantha)
 
-- Learning Rate: 2e-05
-- Steps: 2772
-- Warmup Steps: 277
-- Per Device Train Batch Size: 2
-- Gradient Accumulation Steps 1
-- Optimizer: paged_adamw_8bit
-- Max Sequence Length: 4096
-- Max Prompt Length: 2048
-- Max Length: 2048
-
 ## 💻 Usage
 
 ```python
@@ -43,6 +31,3 @@ pipeline("Hey how are you doing today?")
 
 ```
 
-<div align="center">
-<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/made%20with%20unsloth.png" height="50" align="center" />
-</div>