Update app.py

app.py

@@ -6,14 +6,16 @@ from pynvml import *
 nvmlInit()
 gpu_h = nvmlDeviceGetHandleByIndex(0)
 ctx_limit = 3000
-title = "EagleX 1.7T Demo"
+
+title = "EagleX 2.25T Demo"
+description = f"This is [EagleX 7B 2.25T model](https://blog.rwkv.com/p/336f47bf-d8e9-4174-ac1d-02c6c8a99bc0) - based on the RWKV architecture a 100% attention-free RNN [RWKV-LM](https://wiki.rwkv.com). Supports 100+ world languages and code. And we have [200+ Github RWKV projects](https://github.com/search?o=desc&p=1&q=rwkv&s=updated&type=Repositories). *** Please try examples first (bottom of page) *** (edit them to use your question). Demo limited to context length of {ctx_limit}, download and run locally to run past context length limit"
 
 os.environ["RWKV_JIT_ON"] = '1'
 os.environ["RWKV_CUDA_ON"] = '1' # if '1' then use CUDA kernel for seq mode (much faster)
 
 from rwkv.model import RWKV
-model_path = hf_hub_download(repo_id="recursal/EagleX_1-7T", filename="EagleX-1_7T.pth")
-model = RWKV(model=model_path, strategy='cuda bf16')
+model_path = hf_hub_download(repo_id="RWKV/v5-EagleX-v2-7B-pth", filename=f"v5-EagleX-v2-7B.pth")
+model = RWKV(model=model_path, strategy='cuda fp16i8 *8 -> cuda fp16')
 from rwkv.utils import PIPELINE, PIPELINE_ARGS
 pipeline = PIPELINE(model, "rwkv_vocab_v20230424")
 
@@ -21,11 +23,11 @@ def generate_prompt(instruction, input=""):
     instruction = instruction.strip().replace('\r\n','\n').replace('\n\n','\n')
     input = input.strip().replace('\r\n','\n').replace('\n\n','\n')
     if input:
-        return f"""### Instruction: {instruction}
+        return f"""Instruction: {instruction}
 
-### Input: {input}
+Input: {input}
 
-### Response:"""
+Response:"""
     else:
         return f"""User: hi
 
@@ -86,13 +88,50 @@ def evaluate(
     yield out_str.strip()
 
 examples = [
-    ["### Assistant:\nSure! Here is a very detailed plan to create flying pigs:", 333, 1, 0.3, 0, 1],
-    ["### Assistant:\nSure! Here are some ideas for FTL drive:", 333, 1, 0.3, 0, 1],
-    ["A few light taps upon the pane made her turn to the window. It had begun to snow again.", 333, 1, 0.3, 0, 1],
-    [generate_prompt("Écrivez un programme Python pour miner 1 Bitcoin, avec des commentaires."), 333, 1, 0.3, 0, 1],
+    ["Assistant: How can we craft an engaging story featuring vampires on Mars? Let's think step by step and provide an expert response.", 333, 1, 0.3, 0, 1],
+    ["Assistant: How can we persuade Elon Musk to follow you on Twitter? Let's think step by step and provide an expert response.", 333, 1, 0.3, 0, 1],
+    [generate_prompt("Write a simple website in HTML. When a user clicks the button, it shows a random joke from a list of 4 jokes."), 333, 1, 0.3, 0, 1],
     [generate_prompt("東京で訪れるべき素晴らしい場所とその紹介をいくつか挙げてください。"), 333, 1, 0.3, 0, 1],
     [generate_prompt("Write a story using the following information.", "A man named Alex chops a tree down."), 333, 1, 0.3, 0, 1],
-    ["### Assistant:\nHere is a very detailed plan to kill all mosquitoes:", 333, 1, 0.3, 0, 1]
+    ['''Instruction: You are an expert assistant for summarizing and extracting insights from sales call transcripts. Generate a valid JSON in the following format:
+{
+    "summary": "Summary of the call transcript. ",
+    "products": ["product 1", "product 2"],
+    "rep_name": "Name of the sales rep",
+    "prospect_name": "Name of the prospect",
+    "action_items": ["action item 1", "action item 2"],
+}
+
+Input: John: Hello, this is John.
+Sarah: Hi John, this is Sarah from XYZ Company. I'm calling to discuss our new product, the XYZ Widget, and see if it might be a good fit for your business.
+John: Hi Sarah, thanks for reaching out. I'm definitely interested in learning more about the XYZ Widget. Can you give me a quick overview of what it does?
+Sarah: Of course! The XYZ Widget is a cutting-edge tool that helps businesses streamline their workflow and improve productivity.
+John: That sounds really interesting. I can see how that could benefit our team. Do you have any case studies or success stories from other companies who have used the XYZ Widget?
+Sarah: Absolutely, we have several case studies that I can share with you. I'll send those over along with some additional information about the product. I'd also love to schedule a demo for you and your team to see the XYZ Widget in action.
+John: That would be great. I'll make sure to review the case studies and then we can set up a time for the demo. In the meantime, are there any specific action items or next steps we should take?
+Sarah: Yes, I'll send over the information and then follow up with you to schedule the demo. In the meantime, feel free to reach out if you have any questions or need further information.
+John: Sounds good, I appreciate your help Sarah. I'm looking forward to learning more about the XYZ Widget and seeing how it can benefit our business.
+Sarah: Thank you, John.
+John: You too, bye.
+
+Response:''', 333, 1, 0, 0, 0],
+    ["A few light taps upon the pane made her turn to the window. It had begun to snow again.", 333, 1, 0.3, 0, 1],
+    ['''Edward: I am Edward Elric from Fullmetal Alchemist.
+
+User: Hello Edward. What have you been up to recently?
+
+Edward:''', 333, 1, 0.3, 0, 1],
+    ['''Japanese: 春の初め、桜の花が満開になる頃、小さな町の片隅にある古びた神社の境内は、特別な雰囲気に包まれていた。
+
+English:''', 333, 1, 0.3, 0, 1],
+    ["En una pequeña aldea escondida entre las montañas de Andalucía, donde las calles aún conservaban el eco de antiguas leyendas, vivía un joven llamado Alejandro.", 333, 1, 0.3, 0, 1],
+    ["Dans le cœur battant de Paris, sous le ciel teinté d'un crépuscule d'or et de pourpre, se tenait une petite librairie oubliée par le temps.", 333, 1, 0.3, 0, 1],
+    ["في تطور مذهل وغير مسبوق، أعلنت السلطات المحلية في العاصمة عن اكتشاف أثري قد يغير مجرى التاريخ كما نعرفه.", 333, 1, 0.3, 0, 1],
+    ['''“当然可以，大宇宙不会因为这五公斤就不坍缩了。”关一帆说，他还有一个没说出来的想法：也许大宇宙真的会因为相差一个原子的质量而由封闭转为开放。大自然的精巧有时超出想象，比如生命的诞生，就需要各项宇宙参数在几亿亿分之一精度上的精确配合。但程心仍然可以留下她的生态球，因为在那无数文明创造的无数小宇宙中，肯定有相当一部分不响应回归运动的号召，所以，大宇宙最终被夺走的质量至少有几亿吨，甚至可能是几亿亿亿吨。
+但愿大宇宙能够忽略这个误差。
+程心和关一帆进入了飞船，智子最后也进来了。她早就不再穿那身华丽的和服了，她现在身着迷彩服，再次成为一名轻捷精悍的战士，她的身上佩带着许多武器和生存装备，最引人注目的是那把插在背后的武士刀。
+“放心，我在，你们就在！”智子对两位人类朋友说。
+聚变发动机启动了，推进器发出幽幽的蓝光，''', 333, 1, 0.3, 0, 1],
 ]
 
 ##########################################################################
@@ -100,11 +139,11 @@ examples = [
 with gr.Blocks(title=title) as demo:
     gr.HTML(f"<div style=\"text-align: center;\">\n<h1>RWKV-5 World v2 - {title}</h1>\n</div>")
     with gr.Tab("Raw Generation"):
-        gr.Markdown(f"This is [EagleX 7B 1.7T model](https://substack.recursal.ai/p/eaglex-17t-soaring-past-llama-7b) - based on the RWKV architecture a 100% attention-free RNN [RWKV-LM](https://wiki.rwkv.com). Supports all 100+ world languages and code. And we have [200+ Github RWKV projects](https://github.com/search?o=desc&p=1&q=rwkv&s=updated&type=Repositories). *** Please try examples first (bottom of page) *** (edit them to use your question). Demo limited to ctxlen {ctx_limit}.")
+        gr.Markdown(description)
         with gr.Row():
             with gr.Column():
-                prompt = gr.Textbox(lines=2, label="Prompt", value="### Assistant: Sure! Here is a very detailed plan to create flying pigs:")
-                token_count = gr.Slider(10, 1024, label="Max Tokens", step=10, value=1024)
+                prompt = gr.Textbox(lines=2, label="Prompt", value="Assistant: How can we craft an engaging story featuring vampires on Mars? Let's think step by step and provide an expert response.")
+                token_count = gr.Slider(10, 333, label="Max Tokens", step=10, value=333)
                 temperature = gr.Slider(0.2, 2.0, label="Temperature", step=0.1, value=1.0)
                 top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.3)
                 presence_penalty = gr.Slider(0.0, 1.0, label="Presence Penalty", step=0.1, value=0)
@@ -114,7 +153,7 @@ with gr.Blocks(title=title) as demo:
                     submit = gr.Button("Submit", variant="primary")
                     clear = gr.Button("Clear", variant="secondary")
                 output = gr.Textbox(label="Output", lines=5)
-        data = gr.Dataset(components=[prompt, token_count, temperature, top_p, presence_penalty, count_penalty], samples=examples, label="Example Instructions", headers=["Prompt", "Max Tokens", "Temperature", "Top P", "Presence Penalty", "Count Penalty"])
+        data = gr.Dataset(components=[prompt, token_count, temperature, top_p, presence_penalty, count_penalty], samples=examples, samples_per_page=50, label="Example Instructions", headers=["Prompt", "Max Tokens", "Temperature", "Top P", "Presence Penalty", "Count Penalty"])
         submit.click(evaluate, [prompt, token_count, temperature, top_p, presence_penalty, count_penalty], [output])
         clear.click(lambda: None, [], [output])
         data.click(lambda x: x, [data], [prompt, token_count, temperature, top_p, presence_penalty, count_penalty])