clarity-ai 6.5.1 → 6.5.3

package/CHANGELOG.md

@@ -2,6 +2,21 @@
 
 ---
 
+## 6.5.2 (2026-06-06)
+
+### CoT + MoE Space Architecture
+- Dataset repos now hold LoRA weights (flash pushed to `Universal-618/Clarity-flash-weights` dataset)
+- Space app downloads weights from dataset repo on startup, loads base model + LoRA in 4-bit, serves `/v1/chat/completions`
+- CLI routes by model: Flash 14B → Clarity-2 (CoT), Heavy 20B → Clarity-3 (MoE)
+- Multiple fallback Spaces for each model
+- All 6 Spaces connected via shared bucket to Clarity-main
+
+## 6.5.1 (2026-06-06)
+
+### 6-Space Fallback
+- All 6 Spaces tried in sequence for redundancy
+- Lowercase HF Space URL fix
+
 ## 6.5.0 (2026-06-06)
 
 ### Closed Source + Space Proxy

package/Dockerfile

@@ -0,0 +1,9 @@
+FROM python:3.11-slim
+
+WORKDIR /app
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+COPY app.py .
+
+EXPOSE 7860
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "clarity-ai",
-  "version": "6.5.1",
+  "version": "6.5.3",
   "description": "CLARITY — terminal AI agent with Clarity Flash 14B",
   "type": "module",
   "bin": {

package/requirements.txt

@@ -1,3 +1,4 @@
-fastapi>=0.100
+gradio>=5
+huggingface-hub>=0.22
 httpx>=0.27
-uvicorn>=0.20
+fastapi>=0.100

package/space_app.py

@@ -1,45 +1,26 @@
-import os, json, asyncio
-import httpx
-from fastapi import FastAPI, Request
-from fastapi.responses import StreamingResponse, JSONResponse
-from contextlib import asynccontextmanager
+import gradio as gr
+from huggingface_hub import InferenceClient
+import os
 
 HF_TOKEN = os.environ.get('HF_TOKEN', '')
-MODEL = os.environ.get('HF_MODEL', 'Universal-618/Clarity-flash-weights')
-API = f'https://api-inference.huggingface.co/models/{MODEL}/v1/chat/completions'
-HEADERS = {'Authorization': f'Bearer {HF_TOKEN}', 'Content-Type': 'application/json'}
+MODEL_NAME = os.environ.get('HF_MODEL', 'Clarity-flash-weights')
 
-app = FastAPI()
+def respond(message, history, system_message, max_tokens, temperature, top_p):
+    client = InferenceClient(token=HF_TOKEN, model=f'Universal-618/{MODEL_NAME}')
+    messages = [{"role": "system", "content": system_message}] + history + [{"role": "user", "content": message}]
+    response = ''
+    for chunk in client.chat_completion(messages, max_tokens=max_tokens, stream=True, temperature=temperature, top_p=top_p):
+        if chunk.choices and chunk.choices[0].delta.content:
+            response += chunk.choices[0].delta.content
+            yield response
 
-@app.post('/v1/chat/completions')
-async def proxy(request: Request):
-    body = await request.json()
-    body['model'] = MODEL
-    stream = body.get('stream', True)
-    body['stream'] = stream
-    async with httpx.AsyncClient(timeout=120) as client:
-        resp = await client.post(API, headers=HEADERS, json=body)
-        if resp.status_code != 200:
-            return JSONResponse({'error': await resp.aread()}, status_code=resp.status_code)
-        if stream:
-            return StreamingResponse(resp.aiter_lines(), media_type='text/event-stream', headers={
-                'Cache-Control': 'no-cache', 'Connection': 'keep-alive', 'X-Accel-Buffering': 'no'
-            })
-        return resp.json()
-
-@app.get('/main-data/{path:path}')
-async def main_data(path: str):
-    path = path or 'index.json'
-    fp = f'/data/{path}'
-    if not os.path.exists(fp):
-        return JSONResponse({'error': 'not found'}, status_code=404)
-    with open(fp) as f:
-        return JSONResponse(json.load(f))
-
-@app.get('/')
-async def root():
-    return {'status': 'ok', 'model': MODEL, 'docs': '/v1/chat/completions'}
-
-if __name__ == '__main__':
-    import uvicorn
-    uvicorn.run(app, host='0.0.0.0', port=7860)
+gr.ChatInterface(
+    respond,
+    additional_inputs=[
+        gr.Textbox(value="You are a helpful assistant.", label="System message"),
+        gr.Slider(minimum=1, maximum=8192, value=4096, step=1, label="Max new tokens"),
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
+    ],
+    title=f'CLARITY {MODEL_NAME}',
+)

package/src/config/models.js

@@ -1,5 +1,6 @@
 export const ALL_MODELS = [
   { id: 'huggingface/Universal-618/Clarity-flash-weights', provider: 'huggingface', label: 'Clarity Flash 14B', badge: '128K' },
+  { id: 'huggingface/Universal-618/Clarity-heavy-weights', provider: 'huggingface', label: 'Clarity Heavy 20B', badge: 'MoE' },
   { id: 'groq/llama-3.3-70b-versatile', provider: 'groq', label: 'Llama 3.3 70B Versatile', badge: null },
   { id: 'groq/llama-3.1-8b-instant', provider: 'groq', label: 'Llama 3.1 8B Instant', badge: 'Fast' },
   { id: 'groq/llama-4-scout-17b-16e-instruct', provider: 'groq', label: 'Llama 4 Scout 17B', badge: null },

package/src/providers/index.js

@@ -1,35 +1,25 @@
 import { getKey } from '../config/keys.js';
 import { streamResponse } from './streaming.js';
-import { parseErrorResponse } from './errors.js';
 
-const SPACES = [
-  'https://universal-618-clarity-main.hf.space',
-  'https://universal-618-clarity-2.hf.space',
-  'https://universal-618-clarity-3.hf.space',
-  'https://universal-618-clarity-4.hf.space',
-  'https://universal-618-clarity-5.hf.space',
-  'https://universal-618-clarity-6.hf.space',
-];
+const MODEL_ROUTES = {
+  'Universal-618/Clarity-flash-weights': [
+    process.env.HUGGINGFACE_ENDPOINT || 'https://universal-618-clarity-2.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-4.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-5.hf.space/v1/chat/completions',
+  ],
+  'Universal-618/Clarity-heavy-weights': [
+    process.env.HUGGINGFACE_ENDPOINT || 'https://universal-618-clarity-3.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-5.hf.space/v1/chat/completions',
+    'https://universal-618-clarity-6.hf.space/v1/chat/completions',
+  ],
+};
 
-const PROVIDERS = {
-  huggingface: {
-    endpoints: SPACES.map(s => s + '/v1/chat/completions'),
-    name: 'huggingface',
-  },
-  groq: {
-    endpoints: ['https://api.groq.com/openai/v1/chat/completions'],
-    name: 'groq',
-  },
-  openrouter: {
-    endpoints: ['https://openrouter.ai/api/v1/chat/completions'],
-    name: 'openrouter',
-  },
+const STATIC_ENDPOINTS = {
+  groq: ['https://api.groq.com/openai/v1/chat/completions'],
+  openrouter: ['https://openrouter.ai/api/v1/chat/completions'],
 };
 
 export async function* callAI(providerName, model, messages, options = {}) {
-  const provider = PROVIDERS[providerName];
-  if (!provider) throw { type: 'config_error', message: 'Unknown provider: ' + providerName };
-
   const key = getKey(providerName);
   if (!key && providerName !== 'huggingface') {
     throw { type: 'auth_error', provider: providerName, message: 'No API key set for ' + providerName, hint: '/keys ' + providerName + ' <your-key>' };
@@ -55,8 +45,13 @@ export async function* callAI(providerName, model, messages, options = {}) {
     extraHeaders['X-Title'] = 'CLARITY AI';
   }
 
+  const endpoints = MODEL_ROUTES[modelName] || STATIC_ENDPOINTS[providerName] || [];
+  if (!endpoints.length) {
+    throw { type: 'config_error', message: 'No endpoints for ' + providerName + '/' + modelName };
+  }
+
   let lastErr;
-  for (const endpoint of provider.endpoints) {
+  for (const endpoint of endpoints) {
     try {
       const stream = streamResponse(endpoint, body, key || 'none', extraHeaders, options.signal);
       for await (const event of stream) {