本地运行Colab训练的BERTopic模型触发IndexError问题求助

算力阿峰

2026-8-6

问题

在Colab上训练好BERTopic模型后，本地运行时触发IndexError错误，报错信息如下：

IndexError: Failed in nopython mode pipeline (step: analyzing bytecode)
pop from empty list

使用的代码：

from sentence_transformers import SentenceTransformer
sentence_model = SentenceTransformer('KBLab/sentence-bert-swedish-cased')

model = BERTopic.load('bertopic_model')
text = "my text here for example"
text = [text]

embeddings =  sentence_model.encode(text)
topic, _ = model.transform(text, embeddings)

报错触发在最后一行代码，该代码在Colab运行正常，本地numba及相关库版本已与Colab保持一致，完整报错堆栈：

Traceback (most recent call last):
  File "/home/vaibhav/.local/lib/python3.10/site-packages/flask/app.py", line 2525, in wsgi_app
    response = self.full_dispatch_request()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/flask/app.py", line 1822, in full_dispatch_request
    rv = self.handle_user_exception(e)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/flask/app.py", line 1820, in full_dispatch_request
    rv = self.dispatch_request()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/flask/app.py", line 1796, in dispatch_request
    return self.ensure_sync(self.view_functions[rule.endpoint])(**view_args)
  File "app.py", line 20, in reference_prediction
    preds = data_process(input_api)
  File "data_process.py", line 63, in data_process
    topic, _ = topic_model_mi.transform(text, embeddings)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/bertopic/_bertopic.py", line 423, in transform
    umap_embeddings = self.umap_model.transform(embeddings)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/umap/umap_.py", line 2859, in transform
    dmat = pairwise_distances(
  File "/home/vaibhav/.local/lib/python3.10/site-packages/sklearn/metrics/pairwise.py", line 2022, in pairwise_distances
    return _parallel_pairwise(X, Y, func, n_jobs, **kwds)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/sklearn/metrics/pairwise.py", line 1563, in _parallel_pairwise
    return func(X, Y, **kwds)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/sklearn/metrics/pairwise.py", line 1607, in _pairwise_callable
    out[i, j] = metric(X[i], Y[j], **kwds)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 487, in _compile_for_args
    raise e
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 420, in _compile_for_args
    return_val = self.compile(tuple(argtypes))
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 965, in compile
    cres = self._compiler.compile(args, return_type)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 125, in compile
    status, retval = self._compile_cached(args, return_type)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 139, in _compile_cached
    retval = self._compile_core(args, return_type)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/dispatcher.py", line 152, in _compile_core
    cres = compiler.compile_extra(self.targetdescr.typing_context,
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler.py", line 716, in compile_extra
    return pipeline.compile_extra(func)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler.py", line 452, in compile_extra
    return self._compile_bytecode()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler.py", line 520, in _compile_bytecode
    return self._compile_core()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler.py", line 499, in _compile_core
    raise e
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler.py", line 486, in _compile_core
    pm.run(self.state)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler_machinery.py", line 368, in run
    raise patched_exception
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler_machinery.py", line 356, in run
    self._runPass(idx, pass_inst, state)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler_lock.py", line 35, in _acquire_compile_lock
    return func(*args, **kwargs)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler_machinery.py", line 311, in _runPass
    mutated |= check(pss.run_pass, internal_state)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/compiler_machinery.py", line 273, in check
    mangled = func(compiler_state)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/untyped_passes.py", line 86, in run_pass
    func_ir = interp.interpret(bc)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/interpreter.py", line 1321, in interpret
    flow.run()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/byteflow.py", line 107, in run
    runner.dispatch(state)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/byteflow.py", line 282, in dispatch
    fn(state, inst)
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/byteflow.py", line 1061, in _binaryop
    rhs = state.pop()
  File "/home/vaibhav/.local/lib/python3.10/site-packages/numba/core/byteflow.py", line 1344, in pop
    return self._stack.pop()
IndexError: Failed in nopython mode pipeline (step: analyzing bytecode)
pop from empty list

解决方案

禁用numba的nopython模式：在运行脚本前设置环境变量NUMBA_DISABLE_JIT=1，比如Linux/macOS下执行NUMBA_DISABLE_JIT=1 python your_script.py；或者在代码开头添加：
```
import os
os.environ['NUMBA_DISABLE_JIT'] = '1'
```
强制numba解释执行，绕过JIT编译的兼容性问题。
重新导出/加载模型：在Colab上重新保存模型时使用safetensors序列化方式，代码为model.save("bertopic_model", serialization="safetensors")，之后本地用BERTopic.load("bertopic_model")加载，避免序列化过程中出现的底层代码不兼容。
替换UMAP的距离度量：加载模型后，将UMAP的度量改为通用类型，比如：
```
model.umap_model.metric = "euclidean"
```
部分numba优化的自定义度量在跨环境时易触发错误，改用通用度量可规避。
降级numba版本：尝试安装numba 0.56.4版本，执行pip install numba==0.56.4，该版本对UMAP的兼容性更稳定。

内容的提问来源于stack exchange，提问作者Vai