paper_summary_template.md

title	paper_authors	orgs	paper_link	tags	potm_year	potm_month	paper_order	image_dir	review_author	hidden
<Full Paper Title>	<Forename Surname, et al.' or 'Forename Surname, Forename Surname and Forename Surname.' (max 3 unless >3 first authors)>	<name of org>	https://arxiv.org/abs/<...>	efficient-inference quantisation	<YYYY>	<M>	1	/assets/images/posts/<YYYY-MM>/potm/<short_paper_name>/	name link <Your Name> <e.g. twitter or linkedin url>	true

[200 words is a rough guide for the length of a summary. Feel free to go a fair bit over or under if needs be. The editor will fix any issues with images being rendered too wide/narrow etc. See README for how to view locally if you wish to (not required. Contact CB if this is broken for you.)]

The key idea

A few sentences outlining why the paper is interesting...

Add images where appropriate throughout. This section should always have at least 1 key figure though.

Please use high-res images (zoom in for those screenshots!)

Figure 1a. If the caption isn't included in the image, it should be added like so.

[optional] Background

If necessary, a short intro to background matierial needed to understand the method

Their method

Latex can be included in the standard way, either inline: $R=\sum {t=0}^{\infty }\gamma ^{t}r{t}$

Or as a block:

$$ Q_{t+1}^{A}(s_{t},a_{t})=Q_{t}^{A}(s_{t},a_{t})+\alpha _{t}(s_{t},a_{t})\left(r_{t}+\gamma Q_{t}^{B}\left(s_{t+1},\mathop {\operatorname {arg~max} } _{a}Q_{t}^{A}(s_{t+1},a)\right)-Q_{t}^{A}(s_{t},a_{t})\right). $$

Code can also be included in the standard way:

import popart

builder = popart.Builder()

# Build a simple graph
i1 = builder.addInputTensor(popart.TensorInfo("FLOAT", [1, 2, 32, 32]))
i2 = builder.addInputTensor(popart.TensorInfo("FLOAT", [1, 2, 32, 32]))

o = builder.aiOnnx.add([i1, i2])

Results

...

[optional] Takeaways

...