Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
D
distiller
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Build
Pipelines
Jobs
Pipeline schedules
Artifacts
Deploy
Releases
Package registry
Model registry
Operate
Environments
Terraform modules
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
CI/CD analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
llvm
distiller
Commits
ab02df45
Commit
ab02df45
authored
6 years ago
by
Neta Zmora
Browse files
Options
Downloads
Patches
Plain Diff
Jupyter: new noebook that summarizes the word language model results
parent
b40dff5e
No related branches found
Branches containing commit
No related tags found
Tags containing commit
No related merge requests found
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
jupyter/language_model.ipynb
+117
-0
117 additions, 0 deletions
jupyter/language_model.ipynb
with
117 additions
and
0 deletions
jupyter/language_model.ipynb
0 → 100644
+
117
−
0
View file @
ab02df45
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Word language-model results"
]
},
{
"cell_type": "code",
"execution_count": 66,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 720x360 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"import matplotlib.pyplot as plt\n",
"\n",
"size = [7135600, 28390700, 85917000, 25487550, 16847550, 8591700, 4295850]\n",
"perplexity = [96.29, 84.21, 83.85, 82.93, 83.64, 85.67, 92.79]\n",
"labels = ['small', 'medium', 'large', 'large-70', 'large-80', 'large-90', 'large-95']\n",
"colors = ['r', 'r', 'r', 'b', 'b', 'b', 'b']\n",
"\n",
"fig, ax = plt.subplots(figsize=(10, 5))\n",
"ax.scatter(size, perplexity, c=colors, s=100)\n",
"\n",
"baidu_size = [8591700, 14542053, 17928543, 25190010, 13430535 ]\n",
"baidu_perplexity = [87.70, 85.34, 84.16, 84.05, 83.70]\n",
"baidu_labels = ['large-90', 'large-83', 'large-79', 'large-71', 'large-81.3']\n",
"ax.scatter(baidu_size, baidu_perplexity, c='g', s=100)\n",
"\n",
"\n",
"plt.ylabel('perplexity')\n",
"plt.xlabel('size')\n",
"\n",
"for i, txt in enumerate(labels):\n",
" ax.annotate(txt, (size[i],perplexity[i]+0.4))\n",
" \n",
"for i, txt in enumerate(baidu_labels):\n",
" ax.annotate(txt, (baidu_size[i], baidu_perplexity[i]+0.4))"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f6f2bad42b0>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"import matplotlib.pyplot as plt\n",
"\n",
"size = [7135600, 28390700, 85917000, 25487550, 16847550, 8591700, 4295850]\n",
"perplexity = [96.29, 84.21, 83.85, 82.93, 83.64, 85.67, 92.79]\n",
"labels = ['small', 'medium', 'large', 'large-70', 'large-80', 'large-90', 'large-95']\n",
"colors = ['r', 'r', 'r', 'b', 'b', 'b', 'b']\n",
"\n",
"fig, ax = plt.subplots(figsize=(10, 5))\n",
"ax.scatter(size, perplexity, c=colors, s=100)\n",
"\n",
"\n",
"plt.ylabel('perplexity')\n",
"plt.xlabel('size')\n",
"\n",
"for i, txt in enumerate(labels):\n",
" ax.annotate(txt, (size[i],perplexity[i]+0.4))"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.5.2"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
%% Cell type:markdown id: tags:
## Word language-model results
%% Cell type:code id: tags:
```
python
import
matplotlib.pyplot
as
plt
size
=
[
7135600
,
28390700
,
85917000
,
25487550
,
16847550
,
8591700
,
4295850
]
perplexity
=
[
96.29
,
84.21
,
83.85
,
82.93
,
83.64
,
85.67
,
92.79
]
labels
=
[
'
small
'
,
'
medium
'
,
'
large
'
,
'
large-70
'
,
'
large-80
'
,
'
large-90
'
,
'
large-95
'
]
colors
=
[
'
r
'
,
'
r
'
,
'
r
'
,
'
b
'
,
'
b
'
,
'
b
'
,
'
b
'
]
fig
,
ax
=
plt
.
subplots
(
figsize
=
(
10
,
5
))
ax
.
scatter
(
size
,
perplexity
,
c
=
colors
,
s
=
100
)
baidu_size
=
[
8591700
,
14542053
,
17928543
,
25190010
,
13430535
]
baidu_perplexity
=
[
87.70
,
85.34
,
84.16
,
84.05
,
83.70
]
baidu_labels
=
[
'
large-90
'
,
'
large-83
'
,
'
large-79
'
,
'
large-71
'
,
'
large-81.3
'
]
ax
.
scatter
(
baidu_size
,
baidu_perplexity
,
c
=
'
g
'
,
s
=
100
)
plt
.
ylabel
(
'
perplexity
'
)
plt
.
xlabel
(
'
size
'
)
for
i
,
txt
in
enumerate
(
labels
):
ax
.
annotate
(
txt
,
(
size
[
i
],
perplexity
[
i
]
+
0.4
))
for
i
,
txt
in
enumerate
(
baidu_labels
):
ax
.
annotate
(
txt
,
(
baidu_size
[
i
],
baidu_perplexity
[
i
]
+
0.4
))
```
%% Output
%% Cell type:code id: tags:
```
python
import
matplotlib.pyplot
as
plt
size
=
[
7135600
,
28390700
,
85917000
,
25487550
,
16847550
,
8591700
,
4295850
]
perplexity
=
[
96.29
,
84.21
,
83.85
,
82.93
,
83.64
,
85.67
,
92.79
]
labels
=
[
'
small
'
,
'
medium
'
,
'
large
'
,
'
large-70
'
,
'
large-80
'
,
'
large-90
'
,
'
large-95
'
]
colors
=
[
'
r
'
,
'
r
'
,
'
r
'
,
'
b
'
,
'
b
'
,
'
b
'
,
'
b
'
]
fig
,
ax
=
plt
.
subplots
(
figsize
=
(
10
,
5
))
ax
.
scatter
(
size
,
perplexity
,
c
=
colors
,
s
=
100
)
plt
.
ylabel
(
'
perplexity
'
)
plt
.
xlabel
(
'
size
'
)
for
i
,
txt
in
enumerate
(
labels
):
ax
.
annotate
(
txt
,
(
size
[
i
],
perplexity
[
i
]
+
0.4
))
```
%% Output
%% Cell type:code id: tags:
```
python
```
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment