embeddinggemma-300m Ascend NPU 部署指南

/data/ysws/agentsp/5-15/embeddinggemma-300m-ascend/
├── inference.py          # 推理测试脚本
├── log.txt               # 测试日志
├── README.md             # 本文档
├── test_embeddings.pt    # 测试嵌入结果
├── test_samples.pt       # 测试样本
├── test_samples.json     # 测试样本JSON
└── precision_results.json # 精度测试结果

docker exec -it test-modelagent bash

source /usr/local/Ascend/ascend-toolkit/set_env.sh

cd /data/ysws/agentsp/5-15/embeddinggemma-300m-ascend/

python3 inference.py

python3 inference.py --texts "hello world" "test sentence"

python3 inference.py --device npu:0

cd /data/ysws/agentsp/5-15/embeddinggemma-300m-ascend/

python3 inference.py --precision_test

python3 inference.py --precision_test --num_tests 5

import torch
from transformers import AutoModel, AutoTokenizer

model_path = "/data/ysws/agentsp/5-15/embeddinggemma-300m"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModel.from_pretrained(model_path, trust_remote_code=True)

device = torch.device("npu:0")
model = model.to(device).eval()

texts = ["hello world", "test sentence"]
inputs = tokenizer(texts, padding=True, truncation=True, max_length=128, return_tensors='pt')
inputs = {k: v.to(device) for k, v in inputs.items()}

with torch.no_grad():
    outputs = model(**inputs)

emb = outputs.last_hidden_state.mean(dim=1)
print(f"Embeddings shape: {emb.shape}")
print(f"First embedding: {emb[0][:5].tolist()}")

from sklearn.metrics.pairwise import cosine_similarity
import numpy as np

emb1 = embeddings[0].cpu().numpy()
emb2 = embeddings[1].cpu().numpy()
similarity = cosine_similarity([emb1], [emb2])[0][0]
print(f"Cosine similarity: {similarity}")