skriser
skriser
dssm模型使用subclass方式,在实际预测的时候因为召回要分别取用户的embedding和物品的embedding,那线上推理比如使用模型去推理用户的embedding,在使用subcalss方式中怎么把用户的网络结构拿到呢。我看DeepMatch使用的function API 中使用的是model.__setattr__("user_input", user_inputs_list),model.__setattr__("user_embedding", user_dnn_out)方式。在subclass实现里面有类似方法吗。 感谢
i have dataset: train.csv 5869632 number, user:53424,item:10000; then run error; RuntimeError: CUDA out of memory. Tried to allocate 13.12 GiB (GPU 0; 23.70 GiB total capacity; 0 bytes already allocated;...
# Summary when i use this code ``` faiss.normalize_L2(embeddings) quantizer = faiss.IndexFlatIP(self.dim) index2 = faiss.IndexIVFFlat(quantizer, self.dim, self.nlist, faiss.METRIC_INNER_PRODUCT) index2.add_with_ids(embeddings, ids) ``` I used jmeter for performance test and saved several...
多分类使用时数据量大,无法一次性加载所有数据到内存中,能实现使用生成器模式的功能吗
想增加用户权限校验,使用密码登录后才能看,应该怎么做
执行DSSM深度召回实践.ipynb 报错缺少模块 --------------------------------------------------------------------------- ModuleNotFoundError Traceback (most recent call last) /tmp/ipykernel_1103581/1536832873.py in 7 import torch.nn as nn 8 # from dnn_model import DNN ----> 9 from encoder_model import Encoder 10 import...
--------------------------------------------------------------------------- ModuleNotFoundError Traceback (most recent call last) /tmp/ipykernel_1103581/59926157.py in 6 import torch 7 import torch.nn as nn ----> 8 from dnn_model import DNN 9 from encoder_model import Encoder 10 import...