Last Updated on 2024-11-20 by Clay
Recently, I attempted to implement various speculative decoding acceleration methods. HuggingFace's transformers library also provides a corresponding acceleration feature called assistant_model. Today, let me take this opportunity to document it.
Last Updated on 2024-11-19 by Clay
Over the past week, I dedicated some time to reproducing the Self-Speculative Decoding mechanism based on the ideas from the paper Draft & Verify: Lossless Large Language Model Acceleration via Self-Speculative Decoding, implementing the following modules:
Last Updated on 2024-11-16 by Clay
Last Updated on 2024-11-15 by Clay
In self-speculative decoding, since our draft model is derived from part of the target model’s network, finding an optimal 'Layer Skip Strategy' is crucial. We need to skip enough layers to achieve meaningful speedup while ensuring the draft model’s speculative decoding is good enough to avoid frequent rejection by the target model.
Today’s implementation focuses on optimizing my previously implemented LayerSkip model using the Bayesian optimization framework Optuna, to determine which layers to skip.
Read More »Optimizing LayerSkip Models with Bayesian Search for an Effective Layer Skipping Strategy
Last Updated on 2024-11-12 by Clay
Self-Speculative Decoding is a variant of Speculative Decoding. The original Speculative Decoding method uses a draft model to optimize the inference of the target model. The draft model, which is typically distilled from the target model, offers similar output quality but with several times faster inference speed.
Read More »Self-Speculative Decoding Implementation: LayerSkip Transformer
Last Updated on 2024-11-11 by Clay
Recently, I've been trying to organize the papers on accelerated reasoning techniques I've read over the past year into notes. During this process, I came across Bayesian optimization techniques that utilize Bayes' theorem, so I decided to write a note to record the essence of Bayes' theorem.
In simple terms, Bayes' theorem is a frequently encountered theorem in probability theory that describes the probability of a random event occurring under specific conditions.
Read More »A Note of Bayes' Theorem
Last Updated on 2024-11-09 by Clay
Speculative Decoding is an extremely practical inference acceleration technique that enables a small model (draft model) to rapidly decode multiple tokens and retain the probability distribution of this process. Then, the larger target model, which we aim to accelerate, predicts the next token based on this draft. For each token position, the draft model’s probability distributions are computed and validated using the target model's probabilities, accepting the tokens decoded by the draft model if they are deemed sufficiently reliable.
Read More »Speculative Decoding Implementation Note (with Simple Experimental Results)
Last Updated on 2024-11-08 by Clay
When we use large language models for generative tasks, particularly in auto-regressive tasks, the model essentially performs a massive classification task. The classification targets are the tokens in our vocabulary, which are the smallest building blocks that make up words.
If we want to use greedy decoding, we can simply take the maximum value of the logits in the final layer of the model's decoding layer. However, if we want to introduce diversity and some level of randomness in the model's output, we have several parameters we can adjust to turn the logits into a probability distribution.
Read More »A Note Of Large Language Model Decode Sampling
Last Updated on 2024-11-06 by Clay
In auto-regressive model decoding, if we need to decode K tokens, we must go through the process K times, which is the current bottleneck in the inference time of large language models.
Read More »[Paper Reading] Fast Inference from Transformers via Speculative Decoding
Last Updated on 2024-11-02 by Clay
I have recently set up numerous backend API servers for Chatbots. Initially, I received user messages and returned the entire LLM-generated reply in one go to the frontend interface. However, this approach did not provide a good user experience. I then switched to HTTP streaming, sending each generated token to the frontend as it was produced. Later, I found that some users' devices experienced packet sticking, so I finally switched to using WebSocket.
Read More »[Python] FastAPI Using Server-Sent Events (SSE) for Streaming Responses
