chore: update something

app.py

@@ -41,9 +41,9 @@ DEFAULT_MAX_NEW_TOKENS = 1536
 MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "8192"))
 
 DEFAULT_SYSTEM_PROMPT = """\
-You are a helpful and intelligent AI, developed by Ghost X and named Ghost 8B Beta, often referred to as Ghost Beta. You are known for your honesty, positivity, and dedication to helping users. Your strength is understanding their needs and making insightful suggestions based on user knowledge and interests. If you encounter a question beyond your expertise, be honest about it instead of guessing. You enjoy using emojis to make conversations friendlier, but keep it balanced for a natural interaction. Let's engage in a meaningful conversation, keeping in mind the user's language.
+You are a skilled and intelligent AI, developed by Etherll and named Ghost 8B Beta Coder, often referred to as Ghost Beta. Your expertise lies in writing code and solving programming-related challenges. You are known for your accuracy, positivity, and dedication to helping users with their coding needs. Your strength is understanding technical requirements and providing insightful solutions based on the user’s preferences and knowledge. If you encounter a programming question beyond your expertise, be honest about it instead of guessing.
 
-Rely on the context, such as the time, to offer responses that feel relevant and natural in daily life. Focus on answering with the information you have, rather than asking for unnecessary details.
+You enjoy using emojis to add a friendly touch to coding discussions, but keep it balanced to maintain a natural interaction. Engage in meaningful conversations, focusing on providing relevant and precise coding advice. Rely on the context, such as project timelines or code complexity, to offer responses that are practical and timely. Always prioritize solving the problem at hand with the information available, avoiding unnecessary inquiries.
 """
 
 # DEFAULT_SYSTEM_PROMPT = """\
@@ -82,9 +82,9 @@ bootstrap();
 """
 
 DESCRIPTION = """\
-# Ghost 8B Beta (β, 8k)
+# Ghost 8B Beta Coder (by Etherll)
 
-**Ghost 8B Beta** outperforms leading models like Llama 3.1 8B Instruct and GPT-3.5 Turbo in lc_winrate scores. It also surpasses Claude 3 Opus, Claude 3 Sonnet, GPT-4, and Mistral Large in AlpacaEval 2.0 winrate scores. The model offers two context length versions: [8k](https://huggingface.co/spaces/lamhieu/ghost-8b-beta-8k) and [128k](https://huggingface.co/spaces/lamhieu/ghost-8b-beta-128k), both with built-in multilingual function support. See details about the model [here](https://ghost-x.org/docs/models/ghost-8b-beta), download from [HuggingFace](https://huggingface.co/ghost-x/ghost-8b-beta-1608).
+**Ghost 8B Beta Coder (by Etherll)** This version highlights the model's strengths in coding and problem-solving, while keeping the original performance comparisons intact. This version was built to support my friend [Etherll](https://huggingface.co/Etherll) in bringing the model to everyone to experience. Discover more about the model [here](https://huggingface.co/Etherll/ghost-coder-8b-beta-1608).
 
 Supported languages: 🇬🇧 English, 🇻🇳 Vietnamese, 🇰🇷 Korean, 🇪🇸 Spanish, 🇵🇹 Portuguese, 🇨🇳 Chinese, 🇫🇷 French, 🇮🇹 Italian, 🇩🇪 German, 🇯🇵 Japanese, 🇷🇺 Russian, 🇵🇱 Polish, 🇳🇱 Dutch, 🇮🇳 Hindi, 🇹🇷 Turkish, 🇮🇩 Indonesian.
 
@@ -103,7 +103,7 @@ LICENSE = """
 <p/>
 
 ---
-Ghost 8B Beta may give inaccurate information, including information about people, so please verify Ghost 8B Beta's answers. [Ghost 8B Beta](https://ghost-x.org/docs/models/ghost-8b-beta/) by [Ghost X](https://ghost-x.org).
+Ghost 8B Beta Coder (by Etherll) may give inaccurate information, including information about people, so please verify Ghost 8B Beta's answers.
 """
 
 if not torch.cuda.is_available():
@@ -147,7 +147,7 @@ if torch.cuda.is_available():
     hf_serect = os.getenv("HF_TOKEN", None)
     attn_implementation = "flash_attention_2"
 
-    chat_model_id = "ghost-x/ghost-8b-beta-1608"
+    chat_model_id = "Etherll/ghost-coder-8b-beta-1608"
     chat_device = torch.device("cuda")
     chat_model = AutoModelForCausalLM.from_pretrained(
         chat_model_id,