gcw_6a3w2yXz/deepcpgdna-hou2016-hepg2-NPU
Model Introduction
Files & versions
Pull Requests
Discussion
Analysis

multimolecule/deepcpgdna-hou2016-hepg2 on Ascend NPU

模型与任务

  • 模型:multimolecule/deepcpgdna-hou2016-hepg2
  • Revision:6671f9ba39a3e3e2604986e691aabde45b2953c4
  • 适配路线:regulatory-sequence
  • 设备:npu:0

环境依赖

{
  "torch": "2.9.0+cpu",
  "torch_npu": "available",
  "device_name": "Ascend910_9362",
  "device_count": 2,
  "transformers": "4.57.6",
  "multimolecule": "0.2.1",
  "model_dir": "/tmp/deepcpgdna-hepg2",
  "weight_bytes": 35466461
}

安装:pip install -r requirements.txt

NPU 推理

默认执行:python inference.py

真实输入

{
  "sequence_length": 1001,
  "alphabet": "A/C/G/T/N",
  "pad_token": "N",
  "pad_id": 4,
  "tokenizer": "DnaTokenizer",
  "seq1_length": 184,
  "seq1_sha256": "7a04b9f182884881e3dc7870767e85cf13f1ef93901512f927ca48ef8271903d",
  "seq1_preview": "ACTCCCCTGCCCTCAACAAGATGTTTTGCCAACTGGCCAAGACCTGCCCTGTGCAGCTGTGGGTTGATTCCACACCCCCG...",
  "seq2_length": 331,
  "seq2_sha256": "6eb7f6573c71495bff9eba33a602035a561c556d26fa4b6142fffa96e3bf530f",
  "seq2_preview": "CGCGGGGGTGGCCGGGGCCAGGGCTTCCCACGTGCGCAGCAGGACGCAGCGCTGCCTGAAACTCGCGCCGCGAGGAGAGG...",
  "seq_source": "model card example tumor protein p53 & TERT - verified DNA 5'->3'",
  "encoding": "padding='max_length' max_length=1001 truncation=True",
  "batch": 1,
  "example_input_ids_shape": [
    1,
    1001
  ],
  "example_input_ids_first20": [
    0,
    1,
    3,
    1,
    1,
    1,
    1,
    3,
    2,
    1,
    1,
    1,
    3,
    1,
    0,
    0,
    1,
    0,
    0,
    2
  ],
  "problem_type": "binary per-cell methylation",
  "num_labels": 6,
  "id2label": {
    "0": "HepG21",
    "1": "HepG22",
    "2": "HepG23",
    "3": "HepG24",
    "4": "HepG25",
    "5": "HepG26"
  }
}

真实推理结果

{
  "logits_shape": [
    1,
    6
  ],
  "logits_dtype": "float32",
  "logits_device_cpu": "cpu",
  "logits_device_npu": "npu:0",
  "probs_mean": 0.341821,
  "probs_max": 0.368496,
  "probs_min": 0.325084,
  "logits_cpu_first5": [
    -0.6612699627876282,
    -0.6931700110435486,
    -0.6131107807159424,
    -0.5386757254600525,
    -0.698195219039917
  ],
  "logits_npu_first5": [
    -0.6612597703933716,
    -0.6931613087654114,
    -0.6131016612052917,
    -0.5386651754379272,
    -0.6981855630874634
  ],
  "top3_idx": [
    3,
    2,
    0
  ],
  "top3_labels": [
    "HepG24",
    "HepG23",
    "HepG21"
  ],
  "top3_probs": [
    0.3684956729412079,
    0.3513498902320862,
    0.3404543697834015
  ],
  "finite_ratio": 1.0
}

CPU-NPU 一致性

{
  "logits_max_abs_diff": 1.096725e-05,
  "logits_mean_abs_diff": 9.864569e-06,
  "probs_max_abs_diff": 2.473593e-06,
  "atol": 0.0001,
  "rtol": 0.0001,
  "allclose": true,
  "shape": [
    1,
    6
  ],
  "comparison": "CPU vs NPU same weights, same input, float32"
}

性能

{
  "first_compile_latency_s": 0.2171,
  "stable_avg_latency_s_bs1": 0.0011,
  "sequence_per_s_bs1": 928.91,
  "bases_per_s_bs1": 929838.17,
  "stable_avg_latency_s_bs2": 0.0012,
  "sequence_per_s_bs2": 1715.64,
  "sequence_length": 1001,
  "num_labels": 6,
  "dtype": "float32",
  "warmup_iters": 3,
  "stable_iters": 10,
  "peak_npu_memory_MB": 45.47,
  "device": "npu:0",
  "synchronize": "torch.npu.synchronize() before/after timing"
}

自验证截图

Agent workflow

NPU device call

Model result

三张图片均由 xterm.js 根据本次真实日志生成。

已知限制

  • 仅对本次记录的模型 revision、输入契约、dtype 与 Ascend 环境完成验证。

标签

#NPU #Ascend #Ascend910