DsPy crash course - optimize your LLM pipelines with DsPy (Part 1)
ฝัง
- เผยแพร่เมื่อ 27 มิ.ย. 2024
- We have well-established frameworks like LangChain and LLlamaIndex for building apps with LLMs. So why another framework like DsPy? How is it different from Langchain and LlamaIndex? Let's answer those questions in this video.
This video is Part 1 which is about the different building blocks of DsPy. Part 2 will be about building an end-to-end pipeline with DsPy. So please stay tuned.
⌚️ ⌚️ ⌚️ TIMESTAMPS ⌚️ ⌚️ ⌚️
0:00 - Intro
1:24 - What is DsPy?
3:12 - PyTorch analogy
5:18 - Building blocks of DsPy
5:31 - Language Models
7:47 - Signatures
9:40 - Modules
12:00 - Data
13:01 - Metrics
14:25 - Optimizers
16:04 - Optimization Algorithms
18:02 - Compiling DsPy Module
RELATED LINKS
DsPy Docs - dspy-docs.vercel.app/docs/intro
DsPy GitHub - github.com/stanfordnlp/dspy
DsPy paper - arxiv.org/abs/2310.03714
MY KEY LINKS
TH-cam: / @aibites
Twitter: / ai_bites
Patreon: / ai_bites
Github: github.com/ai-bites
WHO AM I?
I am a Machine Learning researcher/practitioner who has seen the grind of academia and start-ups. I started my career as a software engineer 15 years ago. Because of my love for Mathematics (coupled with a glimmer of luck), I graduated with a Master's in Computer Vision and Robotics in 2016 when the now happening AI revolution started. Life has changed for the better ever since.
#machinelearning #deeplearning #aibites
The best 20m introduction of DSPy so far. Looking forward to part 2.
Cheers.. Comments like this keep me going 🙂
這支影片普遍地解釋了 DSPy 的重要事項。先玩過了之後,有一點經驗但是充滿疑問之時,就可以來看看這支影片,解答諸多疑點。
Really great introduction! Thank you for making this easy to understand.
Glad it was helpful!
Really out of my mind. Never think LLM pipelines can be trained and optimized like traditional ML. But on the second thought, the LLM was trained for general tasks, to use LLM in narrow case such as classification, parsing document, etc. We may need DsPy to optimize for the pipeline (instead of spend time and cost, to fine-tuning or prompt engineering) for more accuracy. Thank you for sharing
exactly! thats what DsPy is trying to achieve. If well received, then we might move away from prompt engineering altogether! 🙂
#ai #llm awesome
Thank you 🙂
audio and video out of sync