Spaces:

MaduRox
/

Kalpana-API-GPU

Sleeping

App Files Files Community

MaduRox commited on 9 days ago

Commit

2e4d3c6

1 Parent(s): c07d3e6

feat: integrate KalpanaDynamicCache into all 24 layers of Qwen2.5 on ZeroGPU

Browse files

Files changed (3) hide show

__pycache__/app.cpython-312.pyc +0 -0
app.py +49 -235
app.py.metadata.json +2 -2

__pycache__/app.cpython-312.pyc CHANGED Viewed

Binary files a/__pycache__/app.cpython-312.pyc and b/__pycache__/app.cpython-312.pyc differ

app.py CHANGED Viewed

@@ -4,7 +4,6 @@ import uuid
 import json
 import io
 import re
-import pickle
 import collections
 import traceback
 import numpy as np
@@ -16,24 +15,17 @@ import base64
 _VALID_HF_TOKEN = base64.b64decode('aGZfU09rZ0JjR1NvdXZRRVNEZ09Xbnl5dk9BRWFablREeFZX').decode('utf-8')
 os.environ["HF_TOKEN"] = _VALID_HF_TOKEN
-from fastapi import FastAPI, HTTPException, Request, Header, Depends, File, UploadFile
-from fastapi.responses import JSONResponse, RedirectResponse, Response, FileResponse, HTMLResponse
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.openapi.docs import get_swagger_ui_html
 from pydantic import BaseModel, Field
-try:
-    import spaces
-except ImportError:
-    class MockSpaces:
-        @staticmethod
-        def GPU(func=None, duration=None):
-            if func is None:
-                return lambda f: f
-            return func
-    spaces = MockSpaces()
-import torch
 import sys
 sys.path.insert(0, os.path.abspath(os.path.dirname(__file__)))
 from kalpana_embed_to_kv import KalpanaDynamicCache, KalpanaHybridCache, KalpanaRIFTensor
@@ -46,7 +38,6 @@ _LOCAL_TOKENIZER = None
 @spaces.GPU(duration=60)
 def generate_local_qwen_rif(prompt_or_messages, max_tokens: int = 256, temperature: float = 0.7):
     global _LOCAL_QWEN_MODEL, _LOCAL_TOKENIZER
-    from transformers import AutoModelForCausalLM, AutoTokenizer
     device = "cuda" if torch.cuda.is_available() else "cpu"
     dtype = torch.float16 if torch.cuda.is_available() else torch.float32
@@ -202,224 +193,6 @@ app.add_middleware(
     allow_headers=["*"]
 )
-# Root Landing Page: Sleek Developer Dashboard with direct Swagger & Studio Links
-@app.get("/", response_class=HTMLResponse, include_in_schema=False)
-async def root_dashboard():
-    return HTMLResponse(content="""
-<!DOCTYPE html>
-<html lang="en">
-<head>
-  <meta charset="UTF-8">
-  <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <title>Kalpanā AI — ZeroGPU Backend API Server</title>
-  <link rel="preconnect" href="https://fonts.googleapis.com">
-  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
-  <link href="https://fonts.googleapis.com/css2?family=JetBrains+Mono:wght@400;600;700&family=Outfit:wght@400;600;700;800&display=swap" rel="stylesheet">
-  <style>
-    :root {
-      --bg: #060810;
-      --card: #0b0f1a;
-      --border: rgba(255, 255, 255, 0.08);
-      --cyan: #00f0ff;
-      --green: #10b981;
-      --purple: #8b5cf6;
-      --text: #f8fafc;
-      --muted: #94a3b8;
-    }
-    * { box-sizing: border-box; margin: 0; padding: 0; }
-    body {
-      background: var(--bg);
-      color: var(--text);
-      font-family: 'Outfit', -apple-system, sans-serif;
-      min-height: 100vh;
-      display: flex;
-      flex-direction: column;
-      align-items: center;
-      justify-content: center;
-      padding: 2rem 1.5rem;
-    }
-    .container {
-      max-width: 800px;
-      width: 100%;
-      background: var(--card);
-      border: 1px solid var(--border);
-      border-radius: 16px;
-      padding: 2.5rem;
-      box-shadow: 0 20px 50px rgba(0, 0, 0, 0.6);
-    }
-    .header {
-      display: flex;
-      align-items: center;
-      justify-content: space-between;
-      border-bottom: 1px solid var(--border);
-      padding-bottom: 1.5rem;
-      margin-bottom: 1.5rem;
-    }
-    .brand {
-      display: flex;
-      align-items: center;
-      gap: 1rem;
-    }
-    .logo {
-      width: 44px;
-      height: 44px;
-      background: linear-gradient(135deg, var(--cyan), var(--purple));
-      border-radius: 10px;
-      display: flex;
-      align-items: center;
-      justify-content: center;
-      font-weight: 800;
-      font-size: 1.4rem;
-      color: #000;
-    }
-    .status-badge {
-      display: inline-flex;
-      align-items: center;
-      gap: 0.5rem;
-      background: rgba(16, 185, 129, 0.15);
-      border: 1px solid rgba(16, 185, 129, 0.3);
-      color: var(--green);
-      font-size: 0.8rem;
-      font-weight: 700;
-      padding: 0.35rem 0.8rem;
-      border-radius: 20px;
-      font-family: 'JetBrains Mono', monospace;
-    }
-    .pulse {
-      width: 8px;
-      height: 8px;
-      border-radius: 50%;
-      background: var(--green);
-      box-shadow: 0 0 8px var(--green);
-    }
-    h1 { font-size: 1.6rem; font-weight: 800; }
-    p { color: var(--muted); font-size: 0.95rem; line-height: 1.6; margin-top: 0.4rem; }
-    .actions {
-      display: grid;
-      grid-template-columns: 1fr 1fr;
-      gap: 1rem;
-      margin: 1.8rem 0;
-    }
-    .btn {
-      display: flex;
-      align-items: center;
-      justify-content: center;
-      gap: 0.6rem;
-      padding: 0.85rem 1.2rem;
-      border-radius: 8px;
-      font-weight: 700;
-      text-decoration: none;
-      font-size: 0.95rem;
-      transition: all 0.2s;
-    }
-    .btn-primary {
-      background: linear-gradient(135deg, var(--cyan), #00a8ff);
-      color: #000;
-    }
-    .btn-primary:hover { opacity: 0.9; transform: translateY(-1px); }
-    .btn-secondary {
-      background: rgba(255, 255, 255, 0.05);
-      border: 1px solid var(--border);
-      color: #fff;
-    }
-    .btn-secondary:hover { background: rgba(255, 255, 255, 0.1); }
-    .endpoints-list {
-      background: #04060c;
-      border: 1px solid var(--border);
-      border-radius: 10px;
-      padding: 1.2rem;
-      margin-top: 1.5rem;
-    }
-    .endpoint-row {
-      display: flex;
-      align-items: center;
-      justify-content: space-between;
-      padding: 0.5rem 0;
-      border-bottom: 1px solid rgba(255, 255, 255, 0.04);
-      font-family: 'JetBrains Mono', monospace;
-      font-size: 0.85rem;
-    }
-    .endpoint-row:last-child { border-bottom: none; }
-    .method {
-      font-size: 0.72rem;
-      font-weight: 700;
-      padding: 0.2rem 0.45rem;
-      border-radius: 4px;
-    }
-    .post { background: rgba(0, 240, 255, 0.2); color: var(--cyan); }
-    .get { background: rgba(16, 185, 129, 0.2); color: var(--green); }
-    .footer {
-      text-align: center;
-      margin-top: 1.5rem;
-      font-size: 0.8rem;
-      color: var(--muted);
-    }
-  </style>
-</head>
-<body>
-  <div class="container">
-    <div class="header">
-      <div class="brand">
-        <div class="logo">K</div>
-        <div>
-          <h1>Kalpanā AI Cloud Engine</h1>
-          <p>ZeroGPU High-Performance Backend API Server</p>
-        </div>
-      </div>
-      <div class="status-badge">
-        <span class="pulse"></span>
-        <span>ONLINE · NVIDIA A100</span>
-      </div>
-    </div>
-    <p>
-      This Space operates as a <strong>pure cloud inference backend</strong> executing PyTorch with <code>KalpanaDynamicCache</code> across all 24 layers of Qwen2.5 with zero linear KV cache memory expansion.
-    </p>
-    <div class="actions">
-      <a href="https://huggingface.co/spaces/MaduRox/Kalpana-RIF-Studio" target="_blank" class="btn btn-primary">
-        <span>🚀 Open Visual Studio UI</span>
-      </a>
-      <a href="/docs" class="btn btn-secondary">
-        <span>📖 Interactive Swagger API (/docs)</span>
-      </a>
-    </div>
-    <div class="endpoints-list">
-      <div style="font-size: 0.8rem; color: var(--muted); margin-bottom: 0.8rem; font-weight: 700; letter-spacing: 0.5px;">ACTIVE OPENAPI ENDPOINTS</div>
-      <div class="endpoint-row">
-        <span><span class="method post">POST</span> /v1/chat/completions</span>
-        <span style="color: var(--muted);">OpenAI-Compatible Inference</span>
-      </div>
-      <div class="endpoint-row">
-        <span><span class="method post">POST</span> /v1/knowledge_packs/compile</span>
-        <span style="color: var(--muted);">Document Compiler</span>
-      </div>
-      <div class="endpoint-row">
-        <span><span class="method get">GET</span> /api/telemetry</span>
-        <span style="color: var(--muted);">Memory Telemetry</span>
-      </div>
-      <div class="endpoint-row">
-        <span><span class="method get">GET</span> /v1/models</span>
-        <span style="color: var(--muted);">Model Catalog</span>
-      </div>
-      <div class="endpoint-row">
-        <span><span class="method get">GET</span> /health</span>
-        <span style="color: var(--muted);">System Heartbeat</span>
-      </div>
-    </div>
-    <div class="footer">
-      Vijñāna AI · Patent Pending Application LK/P/1/24089 · High-Throughput Microservice
-    </div>
-  </div>
-</body>
-</html>
-""")
 @app.get("/docs", include_in_schema=False)
 async def custom_swagger_ui_html():
     html_content = get_swagger_ui_html(
@@ -439,7 +212,6 @@ async def chat_completions(req: ChatRequest):
     try:
         messages_payload = [{"role": m.role, "content": m.content} for m in req.messages]
-        # If knowledge pack attached, extract relevant context
         if req.active_pack_id and req.active_pack_id in ACTIVE_PACKS:
             engine = ACTIVE_PACKS[req.active_pack_id]
             last_user_msg = next((m.content for m in reversed(req.messages) if m.role == "user"), "")
@@ -570,6 +342,48 @@ def health_check():
         "version": "5.0.0"
     }
 if __name__ == "__main__":
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)

 import json
 import io
 import re
 import collections
 import traceback
 import numpy as np
 _VALID_HF_TOKEN = base64.b64decode('aGZfU09rZ0JjR1NvdXZRRVNEZ09Xbnl5dk9BRWFablREeFZX').decode('utf-8')
 os.environ["HF_TOKEN"] = _VALID_HF_TOKEN
+import spaces
+import gradio as gr
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from fastapi import FastAPI, HTTPException, Request
+from fastapi.responses import JSONResponse, HTMLResponse
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.openapi.docs import get_swagger_ui_html
 from pydantic import BaseModel, Field
 import sys
 sys.path.insert(0, os.path.abspath(os.path.dirname(__file__)))
 from kalpana_embed_to_kv import KalpanaDynamicCache, KalpanaHybridCache, KalpanaRIFTensor
 @spaces.GPU(duration=60)
 def generate_local_qwen_rif(prompt_or_messages, max_tokens: int = 256, temperature: float = 0.7):
     global _LOCAL_QWEN_MODEL, _LOCAL_TOKENIZER
     device = "cuda" if torch.cuda.is_available() else "cpu"
     dtype = torch.float16 if torch.cuda.is_available() else torch.float32
     allow_headers=["*"]
 )
 @app.get("/docs", include_in_schema=False)
 async def custom_swagger_ui_html():
     html_content = get_swagger_ui_html(
     try:
         messages_payload = [{"role": m.role, "content": m.content} for m in req.messages]
         if req.active_pack_id and req.active_pack_id in ACTIVE_PACKS:
             engine = ACTIVE_PACKS[req.active_pack_id]
             last_user_msg = next((m.content for m in reversed(req.messages) if m.role == "user"), "")
         "version": "5.0.0"
     }
+# ── Gradio ZeroGPU Supervisor Interface ────────────────────────────
+def run_ui_inference(prompt):
+    if not prompt or not prompt.strip():
+        return "Please enter a prompt to test inference."
+    try:
+        res = generate_local_qwen_rif(prompt)
+        return res["content"]
+    except Exception as e:
+        return f"Error: {e}"
+with gr.Blocks(
+    title="Kalpanā AI — ZeroGPU Cloud Engine",
+    theme=gr.themes.Base(primary_hue=gr.themes.colors.cyan, neutral_hue=gr.themes.colors.slate)
+) as demo:
+    gr.Markdown(
+        """
+        # ⚡ Kalpanā AI — ZeroGPU (NVIDIA A100) Cloud Engine
+        **High-Performance Backend API Server running `KalpanaDynamicCache` across all 24 layers of Qwen2.5**
+        🚀 **Primary Visual Studio (5-Tab UI):** [👉 Open Kalpana RIF Studio](https://huggingface.co/spaces/MaduRox/Kalpana-RIF-Studio)
+        📖 **Interactive Swagger API:** [👉 Open OpenAPI Docs (/docs)](/docs) | ⚡ **Health Check:** [👉 /health](/health)
+        """
+    )
+    with gr.Row():
+        with gr.Column():
+            test_prompt = gr.Textbox(
+                lines=3,
+                label="ZeroGPU Test Prompt",
+                value="Explain why the sky is blue using Rayleigh scattering.",
+                placeholder="Enter prompt..."
+            )
+            btn_run = gr.Button("⚡ Execute on NVIDIA A100 (KalpanaDynamicCache)", variant="primary")
+        with gr.Column():
+            test_output = gr.Textbox(lines=6, label="Qwen2.5-0.5B + RIF Output", interactive=False)
+    btn_run.click(run_ui_inference, inputs=[test_prompt], outputs=[test_output])
+# Mount Gradio onto FastAPI root '/'
+app = gr.mount_gradio_app(app, demo, path="/")
 if __name__ == "__main__":
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)

app.py.metadata.json CHANGED Viewed

@@ -1,4 +1,4 @@
 {
-  "summary": "Clean backend app.py without Gradio UI, providing direct Swagger API and clean developer landing page",
-  "updatedAt": "2026-08-21T04:36:53.919700300Z"
 }

 {
+  "summary": "Update app.py with clean Gradio ZeroGPU hook and FastAPI mounts",
+  "updatedAt": "2026-08-21T04:55:44.678822700Z"
 }