Orangepi AiPro 20T ms is not a valid TensorType

[Terminator98]

I have this setup
Board: OrangePi AiPro 20T Ascend 310B
CANN 8.2.RC1
OS: Ubuntu 22.04
Python 3.10
Mindspore 0.5.0rc1
Mindnlp 2.7.0

I'm running this code below

import mindspore as ms
from mindnlp.transformers import AutoModelForCausalLM, AutoTokenizer

# Suppress harmless NumPy warnings
import warnings
warnings.filterwarnings("ignore", message="The value of the smallest subnormal*")

# Set Ascend context
ms.set_device(device_target="Ascend", device_id=0)

model_name = "Qwen/Qwen2.5-0.5B-Instruct"

# Load model in float16 (required for Ascend 310B)
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, ms_dtype=ms.float16)

# Prepare chat
chat = [
    {"role": "system", "content": "You are a sassy, wise-cracking robot as imagined by Hollywood circa 1986."},
    {"role": "user", "content": "Hey, can you tell me any fun things to do in New York?"}
]

# Format input using Qwen's chat template
input_text = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(input_text, return_tensors="ms")

# Generate without GenerationConfig (uses defaults)
outputs = model.generate(
    input_ids=inputs["input_ids"],
    max_new_tokens=512,
    do_sample=True,
    top_p=0.9,
    temperature=0.7
)

# Decode output
response_text = tokenizer.decode(outputs[0], skip_special_tokens=False)

# Extract assistant's reply
import re
match = re.search(r"<\|im_start\|>assistant\s*(.*?)<\|im_end\|>", response_text, re.DOTALL)
if match:
    assistant_reply = match.group(1).strip()
else:
    # Fallback extraction
    parts = response_text.split("<|im_start|>assistant")
    if len(parts) > 1:
        assistant_reply = parts[-1].split("<|im_end|>")[0].strip()
    else:
        assistant_reply = response_text

print("\n🤖 Assistant Reply:\n")
print(assistant_reply)

I'm getting this error

Traceback (most recent call last):
  File "/home/HwHiAiUser/mostafa/notebooks/mindnlp_test.py", line 25, in <module>
    inputs = tokenizer(input_text, return_tensors="ms")
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 2911, in __call__
    encodings = self._call_one(text=text, text_pair=text_pair, **all_kwargs)
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 3021, in _call_one
    return self.encode_plus(
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 3096, in encode_plus
    return self._encode_plus(
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_fast.py", line 627, in _encode_plus
    batched_output = self._batch_encode_plus(
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_fast.py", line 601, in _batch_encode_plus
    return BatchEncoding(sanitized_tokens, sanitized_encodings, tensor_type=return_tensors)
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 240, in __init__
    self.convert_to_tensors(tensor_type=tensor_type, prepend_batch_axis=prepend_batch_axis)
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/tokenization_utils_base.py", line 707, in convert_to_tensors
    tensor_type = TensorType(tensor_type)
  File "/usr/lib/python3.10/enum.py", line 385, in __call__
    return cls.__new__(cls, value)
  File "/usr/lib/python3.10/enum.py", line 718, in __new__
    raise exc
  File "/usr/lib/python3.10/enum.py", line 700, in __new__
    result = cls._missing_(value)
  File "/home/HwHiAiUser/.venv/lib/python3.10/site-packages/transformers/utils/generic.py", line 501, in _missing_
    raise ValueError(
ValueError: ms is not a valid TensorType, please select one of ['pt', 'tf', 'np', 'jax', 'mlx']

So how to solve this?

[lvyufeng]

ok 我处理下