-
@minhnn1
-
Hi @leo19941227
I was able to get the code run and do some exploration on ASR. However, the performance is too bad. Now I'm trying to figure out what could be the problem. I think one could be the…
-
I know this is usually unnecessary, but I'd like to help out. Do you have a discord?
I have some ideas about datasets, training that I think could be very useful. I'm also a GDScript veteran at this …
-
# How to add new tokens to huggingface transformers vocabulary
In this short article, you’ll learn how to add new tokens to the vocabulary of a huggingface transformer model.
[https://www.depends-on…
-
所参考的文章：https://zhuanlan.zhihu.com/p/365357615
NLP任务整体上可以分为两个类型：
- 理解型任务
- 生成型任务
NLP的四种下游任务：
- `句子对`分类任务
- `单句子`分类任务
- 问答任务
- 单句子标注任务
### 1. 句子对分类任务
- MNLI
```
多体自然语言推理（Multi-Genre Na…
-
### Is your feature request related to a problem? Please describe.
It's not possible to selectively stream audio sources out of Resonite
### Describe the solution you'd like
The ability to spawn a…
-
## Keyword: out of distribution detection
There is no result
## Keyword: out-of-distribution detection
There is no result
## Keyword: expected calibration error
### Soft Augmentation for Image Class…
-
# General
[从Word Embedding到Bert模型——自然语言处理预训练技术发展史](https://www.jiqizhixin.com/articles/2018-12-10-8)
[CS224n自然语言处理与深度学习笔记汇总](https://zhuanlan.zhihu.com/p/31977759)
[深度nlp科普专栏](https://zhuanlan.zhih…
-
## Problem statement
1. performance bottleneck in knowledge-based VQA due to two-phase architecture consists of knowledge retrieval from external soruces and training question answering task in super…
-
- [ ] [S-LoRA: Serving Thousands of Models From One GPU for Fun and Profit - OpenPipe](https://openpipe.ai/blog/s-lora)
# S-LoRA: Serving Thousands of Models From One GPU for Fun and Profit - OpenPi…