🔧 Fix tokenizer compatibility + add fallback model for HF Spaces

app.py

@@ -16,14 +16,29 @@ import re
 # Initialize model
 print("🚀 Loading Jan v1 model...")
 model_name = "janhq/Jan-v1-4B"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype=torch.bfloat16,
-    device_map="auto",
-    load_in_8bit=True
-)
-print("✅ Jan v1 loaded successfully!")
+
+try:
+    tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        torch_dtype=torch.bfloat16,
+        device_map="auto",
+        load_in_8bit=True,
+        trust_remote_code=True
+    )
+    print("✅ Jan v1 loaded successfully!")
+except Exception as e:
+    print(f"❌ Error loading Jan v1: {e}")
+    print("🔄 Falling back to basic model...")
+    # Fallback to a simpler model that works on HF Spaces
+    model_name = "microsoft/DialoGPT-medium"
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name,
+        torch_dtype=torch.float16,
+        device_map="auto"
+    )
+    print("✅ Fallback model loaded!")
 
 class SimpleWebSearch:
     def __init__(self):

requirements.txt

@@ -1,11 +1,11 @@
 # Jan v1 Research Assistant - Complete requirements
-transformers==4.36.2
-torch==2.1.2
-gradio==4.19.2
-accelerate==0.25.0
-bitsandbytes==0.42.0
-sentencepiece==0.1.99
-beautifulsoup4==4.12.3
-requests==2.31.0
-lxml==5.1.0
-validators==0.22.0
+transformers>=4.40.0
+torch>=2.0.0
+gradio>=4.19.0
+accelerate>=0.25.0
+bitsandbytes>=0.42.0
+sentencepiece>=0.1.99
+beautifulsoup4>=4.12.0
+requests>=2.31.0
+lxml>=5.1.0
+validators>=0.22.0