zh:notes:math_softwares
差别
这里会显示出您选择的修订版和当前版本之间的差别。
| 两侧同时换到之前的修订记录 前一修订版 后一修订版 | 前一修订版 | ||
|
zh:notes:math_softwares [2026/08/23 21:11] pzczxs [Vector Model] |
zh:notes:math_softwares [2026/08/24 10:15] (当前版本) pzczxs [Obtain Embedding from a Vector Model] |
||
|---|---|---|---|
| 行 36: | 行 36: | ||
| ==== Obtain Embedding from a Vector Model ==== | ==== Obtain Embedding from a Vector Model ==== | ||
| - | The qwen3.7-text-embedding model is employed to produce a 256-demensional embedding vector for each word. | + | The //text-embedding-v4// model is employed to produce a 64-demensional embedding vector for each word. |
| <code bash> | <code bash> | ||
| - | > python .\extract_embedding.py --model qwen3.7-text-embedding --dimension 25 --output qwen3.7-embeddings.txt --resume .\math_software.word.vocab | + | > python .\extract_embedding.py --model text-embedding-v4 --dimension 64 --batch-size 10 --output text-embedding-v4-64.txt --resume .\math_software.word.vocab |
| </code> | </code> | ||
| ==== Count One- and Two- Grams for PMI Calculation ==== | ==== Count One- and Two- Grams for PMI Calculation ==== | ||
| 行 102: | 行 102: | ||
| </code> | </code> | ||
| - | ==== qwen3.7-text-embedding ==== | + | ==== text-embedding-v4 ==== |
| <code bash> | <code bash> | ||
| > java -Xmx20g -jar GaussianATCreditModel.jar -sf -D 256 -e resources/qwen3.7-embeddings.zip -n 50 -sc 7 -fb data/math_software/10/math_software -K 10 | > java -Xmx20g -jar GaussianATCreditModel.jar -sf -D 256 -e resources/qwen3.7-embeddings.zip -n 50 -sc 7 -fb data/math_software/10/math_software -K 10 | ||
zh/notes/math_softwares.1787490680.txt.gz · 最后更改: 2026/08/23 21:11 由 pzczxs