Update README.md

README.md

@@ -6,13 +6,13 @@ pipeline_tag: text-generation
 
 📝 [Article](https://www.safurai.com/blog/introducing-safurai-csharp)
 
-<center><img src="https://media.discordapp.net/attachments/1071900237414801528/1165927645469478942/mrciffa_A_cartoon_samurai_wearing_a_black_jacket_as_a_chemistry_d4c17e16-567a-41da-9e0e-2902e93def2c.png?ex=6548a1bc&is=65362cbc&hm=5721b5c15d8f97374212970a7d01f17923ef5015d385230b8ae5542fd2d0df21&=&width=1224&height=1224" width="300"></center>
+<center><img src="https://i.imgur.com/REPqbYM.png" width="300"></center>
 
-This is a [`codellama/CodeLlama-7b-hf`](https://huggingface.co/codellama/CodeLlama-7b-hf) model fine-tuned using QLoRA (4-bit precision) on the [`mlabonne/Evol-Instruct-Python-1k`](https://huggingface.co/datasets/mlabonne/Evol-Instruct-Python-1k).
+This is a [`codellama/CodeLlama-7b-hf`](https://huggingface.co/codellama/CodeLlama-7b-hf) model fine-tuned using QLoRA (4-bit precision)
 
 ## 🔧 Training
 
-It was trained on an  in 1h 11m 44s with the following configuration file:
+It was trained on 2 x NVIDIA A100 PCIe 80GB in 7h 40m with the following configuration file:
 
 ```yaml
 base_model: codellama/CodeLlama-34b-hf
@@ -86,11 +86,15 @@ special_tokens:
   unk_token: "<unk>"
 ```
 
-Here are the loss curves:
+Training loss curve:
 
-![](https://i.imgur.com/zrBq01N.png)
+![](https://i.imgur.com/rp1htuf.png)
 
-It is mainly designed for experimental purposes, not for inference.
+Dataset composition:
+
+![](https://i.imgur.com/kTNXgGX.png)
+
+It is mainly designed for experimental purposes.
 
 [<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
 
@@ -103,8 +107,8 @@ from transformers import AutoTokenizer
 import transformers
 import torch
 
-model = "mlabonne/EvolCodeLlama-7b"
-prompt = "Your csharp request"
+model = "Safurai/Evol-csharp-full"
+prompt = "User: \n {your question} \n Assistant: "
 
 tokenizer = AutoTokenizer.from_pretrained(model)
 pipeline = transformers.pipeline(
@@ -120,7 +124,7 @@ sequences = pipeline(
     top_k=10,
     num_return_sequences=1,
     eos_token_id=tokenizer.eos_token_id,
-    max_length=1000,
+    max_length=1024,
 )
 for seq in sequences:
     print(f"Result: {seq['generated_text']}")