Add smolLM2-135M GRPO and SmolLM2-360M

app.py

@@ -36,9 +36,11 @@ MODELS = {
     # "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_Grpo_Gsm8k":{"repo_id":"prithivMLmods/SmolLM2_135M_Grpo_Gsm8k", "desscription":"SmolLM2_135M_Grpo_Gsm8k"},
     "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"},
     "SmolLM2-360M-Instruct-TaiwanChat": {"repo_id": "Luigi/SmolLM2-360M-Instruct-TaiwanChat", "description": "SmolLM2‑360M Instruct fine-tuned on TaiwanChat"},
+    "SmolLM2-360M-Instruct": {"repo_id": "HuggingFaceTB/SmolLM2-360M-Instruct", "description": "Original SmolLM2‑360M Instruct"},
     "Llama-3.2-Taiwan-3B-Instruct": {"repo_id": "lianghsun/Llama-3.2-Taiwan-3B-Instruct", "description": "Llama-3.2-Taiwan-3B-Instruct"},
     "MiniCPM3-4B": {"repo_id": "openbmb/MiniCPM3-4B", "description": "MiniCPM3-4B"},
     "Qwen2.5-3B-Instruct": {"repo_id": "Qwen/Qwen2.5-3B-Instruct", "description": "Qwen2.5-3B-Instruct"},
@@ -76,7 +78,7 @@ 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
@@ -88,7 +90,7 @@ 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