Spaces:
Sleeping
Sleeping
File size: 5,994 Bytes
bac893c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 |
import os
import streamlit as st
import torch
import pandas as pd
import numpy as np
from datasets import load_dataset, Dataset, load_from_disk
from huggingface_hub import login
from streamlit_agraph import agraph, Node, Edge, Config
from sklearn.manifold import TSNE
@st.cache_data
def load_hf_dataset():
# login to huggingface
login(token=os.environ.get("HF_TOKEN"))
# load from huggingface
roster = pd.DataFrame(load_dataset('MAPS-research/GEMRec-Roster', split='train'))
promptBook = pd.DataFrame(load_dataset('MAPS-research/GEMRec-Metadata', split='train'))
# process dataset
roster = roster[['model_id', 'model_name', 'modelVersion_id', 'modelVersion_name',
'model_download_count']].drop_duplicates().reset_index(drop=True)
# add 'custom_score_weights' column to promptBook if not exist
if 'weighted_score_sum' not in promptBook.columns:
promptBook.loc[:, 'weighted_score_sum'] = 0
# merge roster and promptbook
promptBook = promptBook.merge(roster[['model_id', 'model_name', 'modelVersion_id', 'modelVersion_name', 'model_download_count']],
on=['model_id', 'modelVersion_id'], how='left')
# add column to record current row index
promptBook.loc[:, 'row_idx'] = promptBook.index
return roster, promptBook
@st.cache_data
def calc_tsne(prompt_id):
print('==> loading feats')
feats = {}
for pt in os.listdir('../data/feats'):
if pt.split('.')[-1] == 'pt' and pt.split('.')[0].isdigit():
feats[pt.split('.')[0]] = torch.load(os.path.join('../data/feats', pt))
print('==> applying t-SNE')
# apply t-SNE to entries in each feat in feats to get 2D coordinates
tsne = TSNE(n_components=2, random_state=0)
# for k, v in tqdm(feats.items()):
# feats[k]['tsne'] = tsne.fit_transform(v['all'].numpy())
# prompt_id = '90'
feats[prompt_id]['tsne'] = tsne.fit_transform(feats[prompt_id]['all'].numpy())
feats_df = pd.DataFrame(feats[prompt_id]['tsne'], columns=['x', 'y'])
feats_df['prompt_id'] = prompt_id
keys = []
for k in feats[prompt_id].keys():
if k != 'all' and k != 'tsne':
keys.append(int(k.item()))
feats_df['modelVersion_id'] = keys
return feats_df
# print(feats[prompt_id]['tsne'])
if __name__ == '__main__':
st.set_page_config(layout="wide")
# load dataset
roster, promptBook = load_hf_dataset()
# prompt_id = '20'
with st.sidebar:
st.write('## Select Prompt')
prompts = promptBook['prompt_id'].unique().tolist()
# sort prompts by prompt_id
prompts.sort()
prompt_id = st.selectbox('Select Prompt', prompts, index=0)
physics = st.checkbox('Enable Physics')
feats_df = calc_tsne(str(prompt_id))
# keys = []
# for k in feats[prompt_id].keys():
# if k != 'all' and k != 'tsne':
# keys.append(int(k.item()))
# print(keys)
data = []
for idx in feats_df.index:
modelVersion_id = feats_df.loc[idx, 'modelVersion_id']
image_id = promptBook[(promptBook['modelVersion_id'] == modelVersion_id) & (
promptBook['prompt_id'] == int(prompt_id))].reset_index(drop=True).loc[0, 'image_id']
image_url = f"https://modelcofferbucket.s3-accelerate.amazonaws.com/{image_id}.png"
scale = 50
data.append((feats_df.loc[idx, 'x'] * scale, feats_df.loc[idx, 'y'] * scale, image_url))
image_size = promptBook[(promptBook['image_id'] == image_id)].reset_index(drop=True).loc[0, 'size'].split('x')
nodes = []
edges = []
for d in data:
nodes.append( Node(id=d[2],
# label=str(items.loc[idx, 'model_name']),
size=20,
shape="image",
image=d[2],
x=[d[0]],
y=[d[1]],
fixed=False if physics else True,
color={'background': '#00000', 'border': '#ffffff'},
shadow={'enabled': True, 'color': 'rgba(0,0,0,0.4)', 'size': 10, 'x': 1, 'y': 1},
# borderWidth=1,
# shapeProperties={'useBorderWithImage': True},
)
)
# nodes.append( Node(id="Spiderman",
# label="Peter Parker",
# size=25,
# shape="circularImage",
# image="http://marvel-force-chart.surge.sh/marvel_force_chart_img/top_spiderman.png")
# ) # includes **kwargs
# nodes.append( Node(id="Captain_Marvel",
# label="Carol Danvers",
# fixed=True,
# size=25,
# shape="circularImage",
# image="http://marvel-force-chart.surge.sh/marvel_force_chart_img/top_captainmarvel.png")
# )
# edges.append( Edge(source="Captain_Marvel",
# label="friend_of",
# target="Spiderman",
# length=200,
# # **kwargs
# )
# )
#
config = Config(width='100%',
height=800,
directed=True,
physics=physics,
hierarchical=False,
# **kwargs
)
cols = st.columns([3, 1], gap='large')
with cols[0]:
return_value = agraph(nodes=nodes,
edges=edges,
config=config)
# st.write(return_value)
with cols[1]:
try:
st.image(return_value, use_column_width=True)
except:
st.write('No image selected') |