anton-l HF staff commited on
Commit
04547fa
β€’
1 Parent(s): 582e085

Additional metadata fields!

Browse files
Files changed (2) hide show
  1. app.py +80 -20
  2. requirements.txt +2 -1
app.py CHANGED
@@ -153,8 +153,10 @@ def main():
153
  # EVAL RESULTS
154
  ############################
155
  st.markdown("### Evaluation results")
156
- st.markdown("To edit the metrics, you can either use the YAML editor below, or add new metrics using the handy "
157
- "form under it.")
 
 
158
  if "results" not in metadata["model-index"][0]:
159
  metadata["model-index"][0]["results"] = []
160
 
@@ -166,35 +168,80 @@ def main():
166
  results_yaml = st_ace(value=results_yaml, language="yaml")
167
  metadata["model-index"][0]["results"] = try_parse_yaml(results_yaml)
168
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
169
  with st.form(key="eval_form"):
170
- dataset_name = st.text_input(
171
- label="Full name of the dataset", placeholder="Common Voice 8.0"
172
- )
173
  dataset_path = st.text_input(
174
- label="Dataset path / id", placeholder="mozilla-foundation/common_voice_8_0"
 
 
 
 
 
 
 
175
  )
176
  dataset_config = st.text_input(
177
- label="Dataset config (language). Examples: en, pt-BR, clean",
178
  placeholder="en",
179
  )
180
- metric_name = st.text_input(label="Metric name", placeholder="Test WER (+LM)")
 
 
 
 
 
 
 
 
 
181
  metric2name = {"wer": "Word Error Rate", "cer": "Character Error Rate"}
182
  metric_type = st.selectbox(
183
  label="Metric",
184
  options=["wer", "cer"],
185
  format_func=lambda key: metric2name[key],
186
  )
 
 
 
 
 
187
  metric_value = st.text_input(
188
- label="Metric value (0.0 - 100.0)",
189
  placeholder="12.34",
190
  )
191
- try:
192
- metric_value = float(metric_value)
193
- except ValueError:
194
- st.error(f"Couldn't parse `{metric_value}`. Make sure it's a number from 0.0 to 100.0")
 
 
195
 
196
- submitted = st.form_submit_button("Submit")
197
- if submitted:
 
 
 
 
 
 
 
 
 
 
198
  metric = {
199
  "name": metric_name,
200
  "type": metric_type,
@@ -206,7 +253,10 @@ def main():
206
  existing_dataset = existing_result["dataset"]
207
  if (
208
  existing_dataset["type"] == dataset_path
209
- and existing_dataset["args"] == dataset_config
 
 
 
210
  ):
211
  if "metrics" not in existing_result:
212
  existing_result["metrics"] = []
@@ -223,7 +273,9 @@ def main():
223
  "dataset": {
224
  "name": dataset_name,
225
  "type": dataset_path,
226
- "args": dataset_config,
 
 
227
  },
228
  "metrics": [metric],
229
  }
@@ -238,8 +290,14 @@ def main():
238
  )
239
  results_yaml = st_ace(value=results_yaml, language="yaml")
240
  metadata["model-index"][0]["results"] = try_parse_yaml(results_yaml)
241
- st.success(f"Added the metric for {dataset_path} - {dataset_config}! "
242
- f"Check the result in the YAML editor above.")
 
 
 
 
 
 
243
 
244
  ############################
245
  # FINAL YAML
@@ -247,8 +305,10 @@ def main():
247
  st.markdown("## 3. Copy the generated metadata")
248
  st.markdown(
249
  "Copy the YAML from below and replace the metadata at the top of your model's README.md here: "
250
- f"https://huggingface.co/{model_id}/blob/main/README.md"
251
  )
 
 
252
 
253
  new_yaml = yaml.dump(metadata, sort_keys=False, line_break="\n")
254
  st.markdown(f"```yaml\n---\n{new_yaml}---\n```")
 
153
  # EVAL RESULTS
154
  ############################
155
  st.markdown("### Evaluation results")
156
+ st.markdown(
157
+ "To edit the metrics, you can either use the YAML editor below, or add new metrics using the handy "
158
+ "form under it."
159
+ )
160
  if "results" not in metadata["model-index"][0]:
161
  metadata["model-index"][0]["results"] = []
162
 
 
168
  results_yaml = st_ace(value=results_yaml, language="yaml")
169
  metadata["model-index"][0]["results"] = try_parse_yaml(results_yaml)
170
 
171
+ dataset_path_kwargs = {}
172
+ dataset_name_kwargs = {}
173
+ if (
174
+ len(metadata["model-index"][0]["results"]) > 0
175
+ and "dataset" in metadata["model-index"][0]["results"]
176
+ ):
177
+ if "type" in metadata["model-index"][0]["results"]["dataset"]:
178
+ dataset_path_kwargs["value"] = metadata["model-index"][0]["results"][
179
+ "dataset"
180
+ ]["type"]
181
+ if "name" in metadata["model-index"][0]["results"]["dataset"]:
182
+ dataset_name_kwargs["value"] = metadata["model-index"][0]["results"][
183
+ "dataset"
184
+ ]["type"]
185
+
186
  with st.form(key="eval_form"):
 
 
 
187
  dataset_path = st.text_input(
188
+ label="Dataset path / id",
189
+ placeholder="mozilla-foundation/common_voice_8_0",
190
+ **dataset_path_kwargs,
191
+ )
192
+ dataset_name = st.text_input(
193
+ label="A pretty name for the dataset. Examples: 'Common Voice 9.0 (French)', 'LibriSpeech (clean)'",
194
+ placeholder="Common Voice 8.0 (French)",
195
+ **dataset_name_kwargs,
196
  )
197
  dataset_config = st.text_input(
198
+ label="Dataset configuration. Examples: clean, other, en, pt-BR",
199
  placeholder="en",
200
  )
201
+ dataset_language = st.text_input(
202
+ label="Dataset language. Examples: en, pt-BR",
203
+ value=languages[0],
204
+ placeholder="en",
205
+ )
206
+ dataset_split = st.text_input(
207
+ label="Dataset split. Examples: test, validation",
208
+ value="test",
209
+ placeholder="test",
210
+ )
211
  metric2name = {"wer": "Word Error Rate", "cer": "Character Error Rate"}
212
  metric_type = st.selectbox(
213
  label="Metric",
214
  options=["wer", "cer"],
215
  format_func=lambda key: metric2name[key],
216
  )
217
+ metric_name = st.text_input(
218
+ label="A pretty name for the metric. Example: Test WER (+LM)",
219
+ placeholder="Test WER",
220
+ value="Test WER",
221
+ )
222
  metric_value = st.text_input(
223
+ label="Metric value. Use values in range 0.0 to 100.0.",
224
  placeholder="12.34",
225
  )
226
+ # try:
227
+ # metric_value = float(metric_value)
228
+ # except ValueError:
229
+ # st.error(
230
+ # f"Couldn't parse `{metric_value}`. Make sure it's a number from 0.0 to 100.0"
231
+ # )
232
 
233
+ submitted = st.form_submit_button("Add metric")
234
+ if (
235
+ submitted
236
+ and dataset_name
237
+ and dataset_path
238
+ and dataset_config
239
+ and dataset_split
240
+ and dataset_language
241
+ and metric_name
242
+ and metric_type
243
+ and metric_value
244
+ ):
245
  metric = {
246
  "name": metric_name,
247
  "type": metric_type,
 
253
  existing_dataset = existing_result["dataset"]
254
  if (
255
  existing_dataset["type"] == dataset_path
256
+ and "config" in existing_dataset
257
+ and existing_dataset["config"] == dataset_config
258
+ and "split" in existing_dataset
259
+ and existing_dataset["split"] == dataset_split
260
  ):
261
  if "metrics" not in existing_result:
262
  existing_result["metrics"] = []
 
273
  "dataset": {
274
  "name": dataset_name,
275
  "type": dataset_path,
276
+ "config": dataset_config,
277
+ "split": dataset_split,
278
+ "args": {"language": dataset_language},
279
  },
280
  "metrics": [metric],
281
  }
 
290
  )
291
  results_yaml = st_ace(value=results_yaml, language="yaml")
292
  metadata["model-index"][0]["results"] = try_parse_yaml(results_yaml)
293
+ st.success(
294
+ f"Added the metric for {dataset_path} - {dataset_config}! "
295
+ f"Check the result in the YAML editor above."
296
+ )
297
+ elif submitted:
298
+ st.error(
299
+ f"Make sure that you've filled the whole form before clicking 'Add metric'!"
300
+ )
301
 
302
  ############################
303
  # FINAL YAML
 
305
  st.markdown("## 3. Copy the generated metadata")
306
  st.markdown(
307
  "Copy the YAML from below and replace the metadata at the top of your model's README.md here: "
308
+ f"https://huggingface.co/{model_id}/edit/main/README.md"
309
  )
310
+ st.markdown("For mor info on the metadata schema please refer to "
311
+ "https://raw.githubusercontent.com/huggingface/hub-docs/main/modelcard.md")
312
 
313
  new_yaml = yaml.dump(metadata, sort_keys=False, line_break="\n")
314
  st.markdown(f"```yaml\n---\n{new_yaml}---\n```")
requirements.txt CHANGED
@@ -1,2 +1,3 @@
1
  streamlit-tags
2
- streamlit-ace
 
 
1
  streamlit-tags
2
+ streamlit-ace
3
+ streamlit==1.5