Tensorflow Serving Source Code WalkthroughModel Servers Mainmain.ccflag_list: parsed tf model servers options, includes:
port: Port to listen on for gRPC API
grpc_socket_path: listen to a UNIX socket for gRPC API on the given path
rest_api_port: Port to listen on for HTTP/REST API.
rest_api_num_...
Note: Deep contextualized word representations
Peters et al. - 2018 - Deep contextualized word representations
IntroductionThey introduce a new type of deep contextualised word representation complex characteristics of word use. Their word vectors are learned functions of the internal states of ...
Translation is “AI-hard” challenge: preserving the _meaning_ and the _fluency_ of the text.
STATISTICAL MT
Noisy Channel Model
Language model $P(e)$ -> e -> encoder channel $P(f|e)$ -> f -> decoder $argmax P(e|f)$ -> $\hat{e}$
Two components: language model (LM); translation model ...
EVALUATION AND RE-RANKINGHard to characterise the quality of a system’s results:
a subjective problem
query is not the information need
human judgements: too expensive and slow
Automatic evaluation
Simplify assumption:
retrieval is ad-hoc (no prior knowledge of the user)
effectiveness based on ...
WORD-BASED AND PHRASE-BASED MTWord based model: overly simplistic formulation — only used for _alignment_, not actual _translation_.
Phrase based MT: treats N-grams as translation units, referred to as ‘phrases’.
Phrase-pairs memorise:
common translation fragments
common reordering patterns
...
Reference: David Silver, UCL reinforcement learning, lecture 2; CS 294 Deep Reinforcement Learning, Fall 2017
Markov Process (or Markov Chain)Here we assume that the environment is fully observable, which means “the current state completely characterises the process”.
A Markov process is a tupl...
An Overview inside Distributed Tensorflow WorkflowThis article summarizes my understanding of distributed Tensorflow’s workflow. I’d like divide it into following four parts:
Create a server;
Create a session;
Build a computation graph;
Run a session.
So let’s start from creating a server.
Crea...
Stochastic Evolutionary Game Dynamics
Traulsen, A., & Hauert, C. (2009). Stochastic evolutionary game dynamics. Reviews of Nonlinear Dynamics and Complexity, 2, 25–61.
Game theory and evolution早期的 game theory 基于以下假设：
人类是 rational 的；
同时他们也意识到：互动对象知道自己的决策是 rational 的，反之亦然
因此每个个体都可以基于 utili...
TensorFlow Agents：基于TensorFlow的高效批处理增强学习
原文： Hafner, D., Davidson, J., & Vanhoucke, V. (2017). TensorFlow Agents: Efficient Batched Reinforcement Learning in TensorFlow. ArXiv:1709.02878 [Cs]. Retrieved from http://arxiv.org/abs/1709.02878
谷歌大脑的研究人员开发出了TensorFlow Agents：基于TensorFlow的并行增强学习框...
Social-ecological system in multi-agent environment through deep reinforcement learningIntroductionThere are series of urgent global challenges nowadays, which are caused by the overexploitation of common resources, such as overexploitation of groundwater, petroleum, and mine. Due to those issues...
