autodl运行模型报错

autodl V100 运行HuaTuo-o1,报错

File ~/miniconda3/lib/python3.10/site-packages/transformers/models/qwen2/modeling_qwen2.py:191, in Qwen2Attention.forward(self, hidden_states, position_embeddings, attention_mask, past_key_value, cache_position, **kwargs)
    188     else:
    189         attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
--> 191 attn_output, attn_weights = attention_interface(
    192     self,
    193     query_states,
    194     key_states,
    195     value_states,
    196     attention_mask,
    197     dropout=0.0 if not self.training else self.attention_dropout,
    198     scaling=self.scaling,
    199     sliding_window=sliding_window,  # main diff with Llama
    200     **kwargs,
    201 )
    203 attn_output = attn_output.reshape(*input_shape, -1).contiguous()
    204 attn_output = self.o_proj(attn_output)

File ~/miniconda3/lib/python3.10/site-packages/transformers/integrations/sdpa_attention.py:48, in sdpa_attention_forward(module, query, key, value, attention_mask, dropout, scaling, is_causal, **kwargs)
     45 if is_causal is None:
     46     is_causal = causal_mask is None and query.shape[2] > 1
---> 48 attn_output = torch.nn.functional.scaled_dot_product_attention(
     49     query,
     50     key,
     51     value,
     52     attn_mask=causal_mask,
     53     dropout_p=dropout,
     54     scale=scaling,
     55     is_causal=is_causal,
     56 )
     57 attn_output = attn_output.transpose(1, 2).contiguous()
     59 return attn_output, None

RuntimeError: cutlassF: no kernel found to launch! 

报错了，是因为gpu资源，改用半精度就解决了