add all qwen3 variants

app.py

@@ -26,10 +26,14 @@ cancel_event = threading.Event()
 MODELS = {
     "Taiwan-ELM-1_1B-Instruct": {"repo_id": "liswei/Taiwan-ELM-1_1B-Instruct", "description": "Taiwan-ELM-1_1B-Instruct"},
     "Taiwan-ELM-270M-Instruct": {"repo_id": "liswei/Taiwan-ELM-270M-Instruct", "description": "Taiwan-ELM-270M-Instruct"},
-    "Qwen3-8B": {"repo_id": "Qwen/Qwen3-8B", "description": "Qwen3-8B"},
-    "Qwen3-4B": {"repo_id": "Qwen/Qwen3-4B", "description": "Qwen3-4B"},
-    "Qwen3-1.7B": {"repo_id": "Qwen/Qwen3-1,7B", "description": "Qwen3-1.7B"},
-    "Qwen3-0.6B": {"repo_id": "Qwen/Qwen3-0.6B", "description": "Qwen3-0.6B"},
+    "Qwen3-0.6B":    {"repo_id":"Qwen/Qwen3-0.6B","description":"Dense causal language model with 0.6 B total parameters (0.44 B non-embedding), 28 transformer layers, 16 query heads & 8 KV heads, native 32 768-token context window, dual-mode generation, full multilingual & agentic capabilities."},
+    "Qwen3-1.7B":    {"repo_id":"Qwen/Qwen3-1.7B","description":"Dense causal language model with 1.7 B total parameters (1.4 B non-embedding), 28 layers, 16 query heads & 8 KV heads, 32 768-token context, stronger reasoning vs. 0.6 B variant, dual-mode inference, instruction following across 100+ languages."},
+    "Qwen3-4B":      {"repo_id":"Qwen/Qwen3-4B","description":"Dense causal language model with 4.0 B total parameters (3.6 B non-embedding), 36 layers, 32 query heads & 8 KV heads, native 32 768-token context (extendable to 131 072 via YaRN), balanced mid-range capacity & long-context reasoning."},
+    "Qwen3-8B":      {"repo_id":"Qwen/Qwen3-8B","description":"Dense causal language model with 8.2 B total parameters (6.95 B non-embedding), 36 layers, 32 query heads & 8 KV heads, 32 768-token context (131 072 via YaRN), excels at multilingual instruction following & zero-shot tasks."},
+    "Qwen3-14B":     {"repo_id":"Qwen/Qwen3-14B","description":"Dense causal language model with 14.8 B total parameters (13.2 B non-embedding), 40 layers, 40 query heads & 8 KV heads, 32 768-token context (131 072 via YaRN), enhanced human preference alignment & advanced agent integration."},
+    "Qwen3-32B":     {"repo_id":"Qwen/Qwen3-32B","description":"Dense causal language model with 32.8 B total parameters (31.2 B non-embedding), 64 layers, 64 query heads & 8 KV heads, 32 768-token context (131 072 via YaRN), flagship variant delivering state-of-the-art reasoning & instruction following."},
+    "Qwen3-30B-A3B": {"repo_id":"Qwen/Qwen3-30B-A3B","description":"Mixture-of-Experts model with 30.5 B total parameters (29.9 B non-embedding, 3.3 B activated per token), 48 layers, 128 experts (8 activated per token), 32 query heads & 4 KV heads, 32 768-token context (131 072 via YaRN), MoE routing for scalable specialized reasoning."},
+    "Qwen3-235B-A22B":{"repo_id":"Qwen/Qwen3-235B-A22B","description":"Mixture-of-Experts model with 235 B total parameters (234 B non-embedding, 22 B activated per token), 94 layers, 128 experts (8 activated per token), 64 query heads & 4 KV heads, 32 768-token context (131 072 via YaRN), ultra-scale reasoning & agentic workflows."},
     "Gemma-3-4B-IT": {"repo_id": "unsloth/gemma-3-4b-it", "description": "Gemma-3-4B-IT"},
     "SmolLM2-135M-Instruct-TaiwanChat": {"repo_id": "Luigi/SmolLM2-135M-Instruct-TaiwanChat", "description": "SmolLM2‑135M Instruct fine-tuned on TaiwanChat"},
     "SmolLM2-135M-Instruct": {"repo_id": "HuggingFaceTB/SmolLM2-135M-Instruct", "description": "Original SmolLM2‑135M Instruct"},
@@ -66,7 +70,7 @@ def load_pipeline(model_name):
                 tokenizer=tokenizer,
                 trust_remote_code=True,
                 torch_dtype=dtype,
-                device_map="auto"
+                device_map="xpu"
             )
             PIPELINES[model_name] = pipe
             return pipe
@@ -78,7 +82,7 @@ def load_pipeline(model_name):
         model=repo,
         tokenizer=tokenizer,
         trust_remote_code=True,
-        device_map="auto"
+        device_map="xpu"
     )
     PIPELINES[model_name] = pipe
     return pipe