Alexander Seifert commited on
Commit
2918df9
β€’
1 Parent(s): d19773e

move files

Browse files
README.md CHANGED
@@ -6,7 +6,7 @@ colorTo: indigo
6
  python_version: 3.9
7
  sdk: streamlit
8
  sdk_version: 1.10.0
9
- app_file: main.py
10
  pinned: true
11
  ---
12
 
 
6
  python_version: 3.9
7
  sdk: streamlit
8
  sdk_version: 1.10.0
9
+ app_file: src/main.py
10
  pinned: true
11
  ---
12
 
data.py β†’ src/data.py RENAMED
File without changes
load.py β†’ src/load.py RENAMED
File without changes
main.py β†’ src/main.py RENAMED
File without changes
model.py β†’ src/model.py RENAMED
File without changes
{subpages β†’ src/subpages}/__init__.py RENAMED
File without changes
{subpages β†’ src/subpages}/attention.py RENAMED
@@ -70,8 +70,7 @@ SETUP_HTML = """
70
  JS_TEMPLATE = """requirejs(['basic', 'ecco'], function(basic, ecco){{
71
  const viz_id = basic.init()
72
 
73
- ecco.interactiveTokensAndFactorSparklines(viz_id, {},
74
- {{
75
  'hltrCFG': {{'tokenization_config': {{'token_prefix': '', 'partial_token_prefix': '##'}}
76
  }}
77
  }})
@@ -160,12 +159,7 @@ class AttentionPage(Page):
160
  data = nmf.explore(returnData=True)
161
  JS_TEMPLATE = f"""<script>requirejs(['basic', 'ecco'], function(basic, ecco){{
162
  const viz_id = basic.init()
163
-
164
- ecco.interactiveTokensAndFactorSparklines(viz_id, {data},
165
- {{
166
- 'hltrCFG': {{'tokenization_config': {{'token_prefix': '', 'partial_token_prefix': '##'}}
167
- }}
168
- }})
169
  }}, function (err) {{
170
  console.log(err);
171
  }})</script>"""
 
70
  JS_TEMPLATE = """requirejs(['basic', 'ecco'], function(basic, ecco){{
71
  const viz_id = basic.init()
72
 
73
+ ecco.interactiveTokensAndFactorSparklines(viz_id, {}, {{
 
74
  'hltrCFG': {{'tokenization_config': {{'token_prefix': '', 'partial_token_prefix': '##'}}
75
  }}
76
  }})
 
159
  data = nmf.explore(returnData=True)
160
  JS_TEMPLATE = f"""<script>requirejs(['basic', 'ecco'], function(basic, ecco){{
161
  const viz_id = basic.init()
162
+ ecco.interactiveTokensAndFactorSparklines(viz_id, {data}, {{ 'hltrCFG': {{'tokenization_config': {{'token_prefix': '', 'partial_token_prefix': '##'}} }} }})
 
 
 
 
 
163
  }}, function (err) {{
164
  console.log(err);
165
  }})</script>"""
{subpages β†’ src/subpages}/debug.py RENAMED
File without changes
{subpages β†’ src/subpages}/emoji-en-US.json RENAMED
File without changes
{subpages β†’ src/subpages}/faiss.py RENAMED
File without changes
{subpages β†’ src/subpages}/find_duplicates.py RENAMED
File without changes
{subpages β†’ src/subpages}/hidden_states.py RENAMED
@@ -135,7 +135,7 @@ class HiddenStatesPage(Page):
135
  x="x",
136
  y="y",
137
  color="labels",
138
- hover_data=["sent0", "sent1", "sent2", "sent3", "sent4"],
139
  hover_name="tokens",
140
  title="Colored by label",
141
  )
@@ -147,7 +147,7 @@ class HiddenStatesPage(Page):
147
  x="x",
148
  y="y",
149
  color="preds",
150
- hover_data=["sent0", "sent1", "sent2", "sent3", "sent4"],
151
  hover_name="tokens",
152
  title="Colored by prediction",
153
  )
 
135
  x="x",
136
  y="y",
137
  color="labels",
138
+ hover_data=["ids", "preds", "sent0", "sent1", "sent2", "sent3", "sent4"],
139
  hover_name="tokens",
140
  title="Colored by label",
141
  )
 
147
  x="x",
148
  y="y",
149
  color="preds",
150
+ hover_data=["ids", "labels", "sent0", "sent1", "sent2", "sent3", "sent4"],
151
  hover_name="tokens",
152
  title="Colored by prediction",
153
  )
{subpages β†’ src/subpages}/home.py RENAMED
@@ -6,7 +6,7 @@ import streamlit as st
6
 
7
  from data import get_data
8
  from subpages.page import Context, Page
9
- from utils import classmap, color_map_color
10
 
11
  _SENTENCE_ENCODER_MODEL = (
12
  "sentence-transformers/all-MiniLM-L6-v2",
@@ -138,7 +138,7 @@ class HomePage(Page):
138
  with col2b:
139
  st.subheader("β€”")
140
  st.write("**Icon**")
141
- emojis = list(json.load(open("subpages/emoji-en-US.json")).keys())
142
  for label in labels:
143
  if f"icon_{label}" not in st.session_state:
144
  st.session_state[f"icon_{label}"] = classmap[label]
 
6
 
7
  from data import get_data
8
  from subpages.page import Context, Page
9
+ from utils import PROJ, classmap, color_map_color
10
 
11
  _SENTENCE_ENCODER_MODEL = (
12
  "sentence-transformers/all-MiniLM-L6-v2",
 
138
  with col2b:
139
  st.subheader("β€”")
140
  st.write("**Icon**")
141
+ emojis = list(json.load(open(PROJ / "subpages/emoji-en-US.json")).keys())
142
  for label in labels:
143
  if f"icon_{label}" not in st.session_state:
144
  st.session_state[f"icon_{label}"] = classmap[label]
{subpages β†’ src/subpages}/inspect.py RENAMED
File without changes
{subpages β†’ src/subpages}/losses.py RENAMED
File without changes
{subpages β†’ src/subpages}/lossy_samples.py RENAMED
File without changes
{subpages β†’ src/subpages}/metrics.py RENAMED
File without changes
{subpages β†’ src/subpages}/misclassified.py RENAMED
File without changes
{subpages β†’ src/subpages}/page.py RENAMED
File without changes
{subpages β†’ src/subpages}/probing.py RENAMED
File without changes
{subpages β†’ src/subpages}/random_samples.py RENAMED
File without changes
{subpages β†’ src/subpages}/raw_data.py RENAMED
File without changes
utils.py β†’ src/utils.py RENAMED
@@ -1,3 +1,5 @@
 
 
1
  import matplotlib as matplotlib
2
  import matplotlib.cm as cm
3
  import pandas as pd
@@ -7,6 +9,8 @@ import torch
7
  import torch.nn.functional as F
8
  from st_aggrid import AgGrid, GridOptionsBuilder, GridUpdateMode
9
 
 
 
10
  tokenizer_hash_funcs = {
11
  tokenizers.Tokenizer: lambda _: None,
12
  tokenizers.AddedToken: lambda _: None,
 
1
+ from pathlib import Path
2
+
3
  import matplotlib as matplotlib
4
  import matplotlib.cm as cm
5
  import pandas as pd
 
9
  import torch.nn.functional as F
10
  from st_aggrid import AgGrid, GridOptionsBuilder, GridUpdateMode
11
 
12
+ PROJ = Path(__file__).parent
13
+
14
  tokenizer_hash_funcs = {
15
  tokenizers.Tokenizer: lambda _: None,
16
  tokenizers.AddedToken: lambda _: None,