Spaces:
Running
Running
add model_link
Browse files- app.py +8 -92
- results/Chronos_small/config.json +2 -1
- results/Moirai_base/config.json +2 -1
- results/Moirai_large/config.json +2 -1
- results/Moirai_small/config.json +2 -1
- results/chronos_base/config.json +2 -1
- results/chronos_large/config.json +2 -1
- results/timer_small/config.json +5 -0
- results/timesfm/config.json +2 -1
- src/display/utils.py +3 -3
- src/leaderboard/read_evals.py +1 -1
app.py
CHANGED
|
@@ -110,17 +110,23 @@ def init_leaderboard(ori_dataframe, model_info_df):
|
|
| 110 |
if ori_dataframe is None or ori_dataframe.empty:
|
| 111 |
raise ValueError("Leaderboard DataFrame is empty or None.")
|
| 112 |
model_info_col_list = [c.name for c in fields(ModelInfoColumn) if c.displayed_by_default if c.name not in ['#Params (B)', 'available_on_hub', 'hub', 'Model sha','Hub License']]
|
|
|
|
| 113 |
default_selection_list = list(ori_dataframe.columns) + model_info_col_list
|
| 114 |
-
print('default_selection_list: ', default_selection_list)
|
| 115 |
# ipdb.set_trace()
|
| 116 |
# default_selection_list = [col for col in default_selection_list if col not in ['#Params (B)', 'available_on_hub', 'hub', 'Model sha','Hub License']]
|
| 117 |
merged_df = get_merged_df(ori_dataframe, model_info_df)
|
| 118 |
new_cols = ['T'] + [col for col in merged_df.columns if col != 'T']
|
| 119 |
merged_df = merged_df[new_cols]
|
| 120 |
print('Merged df: ', merged_df)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
return Leaderboard(
|
| 122 |
value=merged_df,
|
| 123 |
-
|
| 124 |
select_columns=SelectColumns(
|
| 125 |
default_selection=default_selection_list,
|
| 126 |
# default_selection=[c.name for c in fields(ModelInfoColumn) if
|
|
@@ -183,96 +189,6 @@ with demo:
|
|
| 183 |
with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=4):
|
| 184 |
gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
|
| 185 |
|
| 186 |
-
# with gr.TabItem("🚀 Submit here! ", elem_id="llm-benchmark-tab-table", id=5):
|
| 187 |
-
# with gr.Column():
|
| 188 |
-
# with gr.Row():
|
| 189 |
-
# gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")
|
| 190 |
-
#
|
| 191 |
-
# with gr.Column():
|
| 192 |
-
# with gr.Accordion(
|
| 193 |
-
# f"✅ Finished Evaluations ({len(finished_eval_queue_df)})",
|
| 194 |
-
# open=False,
|
| 195 |
-
# ):
|
| 196 |
-
# with gr.Row():
|
| 197 |
-
# finished_eval_table = gr.components.Dataframe(
|
| 198 |
-
# value=finished_eval_queue_df,
|
| 199 |
-
# headers=EVAL_COLS,
|
| 200 |
-
# datatype=EVAL_TYPES,
|
| 201 |
-
# row_count=5,
|
| 202 |
-
# )
|
| 203 |
-
# with gr.Accordion(
|
| 204 |
-
# f"🔄 Running Evaluation Queue ({len(running_eval_queue_df)})",
|
| 205 |
-
# open=False,
|
| 206 |
-
# ):
|
| 207 |
-
# with gr.Row():
|
| 208 |
-
# running_eval_table = gr.components.Dataframe(
|
| 209 |
-
# value=running_eval_queue_df,
|
| 210 |
-
# headers=EVAL_COLS,
|
| 211 |
-
# datatype=EVAL_TYPES,
|
| 212 |
-
# row_count=5,
|
| 213 |
-
# )
|
| 214 |
-
#
|
| 215 |
-
# with gr.Accordion(
|
| 216 |
-
# f"⏳ Pending Evaluation Queue ({len(pending_eval_queue_df)})",
|
| 217 |
-
# open=False,
|
| 218 |
-
# ):
|
| 219 |
-
# with gr.Row():
|
| 220 |
-
# pending_eval_table = gr.components.Dataframe(
|
| 221 |
-
# value=pending_eval_queue_df,
|
| 222 |
-
# headers=EVAL_COLS,
|
| 223 |
-
# datatype=EVAL_TYPES,
|
| 224 |
-
# row_count=5,
|
| 225 |
-
# )
|
| 226 |
-
# with gr.Row():
|
| 227 |
-
# gr.Markdown("# ✉️✨ Submit your model outputs !", elem_classes="markdown-text")
|
| 228 |
-
# gr.Markdown(
|
| 229 |
-
# "Send your model outputs for all the models using the ContextualBench code and email them to us at [email protected] ",
|
| 230 |
-
# elem_classes="markdown-text")
|
| 231 |
-
|
| 232 |
-
# with gr.Row():
|
| 233 |
-
# with gr.Column():
|
| 234 |
-
# model_name_textbox = gr.Textbox(label="Model name")
|
| 235 |
-
# revision_name_textbox = gr.Textbox(label="Revision commit", placeholder="main")
|
| 236 |
-
# model_type = gr.Dropdown(
|
| 237 |
-
# choices=[t.to_str(" : ") for t in ModelType if t != ModelType.Unknown],
|
| 238 |
-
# label="Model type",
|
| 239 |
-
# multiselect=False,
|
| 240 |
-
# value=None,
|
| 241 |
-
# interactive=True,
|
| 242 |
-
# )
|
| 243 |
-
|
| 244 |
-
# with gr.Column():
|
| 245 |
-
# precision = gr.Dropdown(
|
| 246 |
-
# choices=[i.value.name for i in Precision if i != Precision.Unknown],
|
| 247 |
-
# label="Precision",
|
| 248 |
-
# multiselect=False,
|
| 249 |
-
# value="float16",
|
| 250 |
-
# interactive=True,
|
| 251 |
-
# )
|
| 252 |
-
# weight_type = gr.Dropdown(
|
| 253 |
-
# choices=[i.value.name for i in WeightType],
|
| 254 |
-
# label="Weights type",
|
| 255 |
-
# multiselect=False,
|
| 256 |
-
# value="Original",
|
| 257 |
-
# interactive=True,
|
| 258 |
-
# )
|
| 259 |
-
# base_model_name_textbox = gr.Textbox(label="Base model (for delta or adapter weights)")
|
| 260 |
-
|
| 261 |
-
# submit_button = gr.Button("Submit Eval")
|
| 262 |
-
# submission_result = gr.Markdown()
|
| 263 |
-
# submit_button.click(
|
| 264 |
-
# add_new_eval,
|
| 265 |
-
# [
|
| 266 |
-
# model_name_textbox,
|
| 267 |
-
# base_model_name_textbox,
|
| 268 |
-
# revision_name_textbox,
|
| 269 |
-
# precision,
|
| 270 |
-
# weight_type,
|
| 271 |
-
# model_type,
|
| 272 |
-
# ],
|
| 273 |
-
# submission_result,
|
| 274 |
-
# )
|
| 275 |
-
|
| 276 |
with gr.Row():
|
| 277 |
with gr.Accordion("📙 Citation", open=False):
|
| 278 |
citation_button = gr.Textbox(
|
|
|
|
| 110 |
if ori_dataframe is None or ori_dataframe.empty:
|
| 111 |
raise ValueError("Leaderboard DataFrame is empty or None.")
|
| 112 |
model_info_col_list = [c.name for c in fields(ModelInfoColumn) if c.displayed_by_default if c.name not in ['#Params (B)', 'available_on_hub', 'hub', 'Model sha','Hub License']]
|
| 113 |
+
col2type_dict = {c.name: c.type for c in fields(ModelInfoColumn)}
|
| 114 |
default_selection_list = list(ori_dataframe.columns) + model_info_col_list
|
| 115 |
+
# print('default_selection_list: ', default_selection_list)
|
| 116 |
# ipdb.set_trace()
|
| 117 |
# default_selection_list = [col for col in default_selection_list if col not in ['#Params (B)', 'available_on_hub', 'hub', 'Model sha','Hub License']]
|
| 118 |
merged_df = get_merged_df(ori_dataframe, model_info_df)
|
| 119 |
new_cols = ['T'] + [col for col in merged_df.columns if col != 'T']
|
| 120 |
merged_df = merged_df[new_cols]
|
| 121 |
print('Merged df: ', merged_df)
|
| 122 |
+
# get the data type
|
| 123 |
+
datatype_list = [col2type_dict[col] if col in col2type_dict else 'number' for col in merged_df.columns]
|
| 124 |
+
# print('datatype_list: ', datatype_list)
|
| 125 |
+
# print('merged_df.column: ', merged_df.columns)
|
| 126 |
+
# ipdb.set_trace()
|
| 127 |
return Leaderboard(
|
| 128 |
value=merged_df,
|
| 129 |
+
datatype=[c.type for c in fields(ModelInfoColumn)],
|
| 130 |
select_columns=SelectColumns(
|
| 131 |
default_selection=default_selection_list,
|
| 132 |
# default_selection=[c.name for c in fields(ModelInfoColumn) if
|
|
|
|
| 189 |
with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=4):
|
| 190 |
gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
|
| 191 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 192 |
with gr.Row():
|
| 193 |
with gr.Accordion("📙 Citation", open=False):
|
| 194 |
citation_button = gr.Textbox(
|
results/Chronos_small/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_small",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_small",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/amazon/chronos-t5-small"
|
| 6 |
}
|
results/Moirai_base/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_base",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_base",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/Salesforce/moirai-1.1-R-base"
|
| 6 |
}
|
results/Moirai_large/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_large",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_large",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/Salesforce/moirai-1.1-R-large"
|
| 6 |
}
|
results/Moirai_small/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_small",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Moirai_small",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/Salesforce/moirai-1.1-R-large"
|
| 6 |
}
|
results/chronos_base/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_base",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_base",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/amazon/chronos-t5-base"
|
| 6 |
}
|
results/chronos_large/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_large",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "Chronos_large",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/amazon/chronos-t5-large"
|
| 6 |
}
|
results/timer_small/config.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": "timer_small",
|
| 3 |
+
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32"
|
| 5 |
+
}
|
results/timesfm/config.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"model": "TimesFM",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
-
"model_dtype": "float32"
|
|
|
|
| 5 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"model": "TimesFM",
|
| 3 |
"model_type": "pretrained",
|
| 4 |
+
"model_dtype": "float32",
|
| 5 |
+
"model_link": "https://huggingface.co/google/timesfm-1.0-200m"
|
| 6 |
}
|
src/display/utils.py
CHANGED
|
@@ -27,14 +27,14 @@ model_info_dict.append(["model_type_symbol", ColumnContent, ColumnContent("T", "
|
|
| 27 |
model_info_dict.append(["model", ColumnContent, ColumnContent("model", "markdown", True, never_hidden=True)])
|
| 28 |
# Model information
|
| 29 |
model_info_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False, True)])
|
| 30 |
-
model_info_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
|
| 31 |
-
model_info_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
|
| 32 |
model_info_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False, True)])
|
| 33 |
model_info_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False, True)])
|
| 34 |
model_info_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False, True)])
|
| 35 |
model_info_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False, True)])
|
| 36 |
model_info_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
|
| 37 |
-
model_info_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
|
| 38 |
|
| 39 |
# We use make dataclass to dynamically fill the scores from Tasks
|
| 40 |
ModelInfoColumn = make_dataclass("ModelInfoColumn", model_info_dict, frozen=True)
|
|
|
|
| 27 |
model_info_dict.append(["model", ColumnContent, ColumnContent("model", "markdown", True, never_hidden=True)])
|
| 28 |
# Model information
|
| 29 |
model_info_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False, True)])
|
| 30 |
+
# model_info_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
|
| 31 |
+
# model_info_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
|
| 32 |
model_info_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False, True)])
|
| 33 |
model_info_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False, True)])
|
| 34 |
model_info_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False, True)])
|
| 35 |
model_info_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False, True)])
|
| 36 |
model_info_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
|
| 37 |
+
# model_info_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
|
| 38 |
|
| 39 |
# We use make dataclass to dynamically fill the scores from Tasks
|
| 40 |
ModelInfoColumn = make_dataclass("ModelInfoColumn", model_info_dict, frozen=True)
|
src/leaderboard/read_evals.py
CHANGED
|
@@ -42,7 +42,7 @@ class ModelConfig:
|
|
| 42 |
def to_dict(self):
|
| 43 |
"""Converts the model info to a dict compatible with our dataframe display"""
|
| 44 |
data_dict = {
|
| 45 |
-
|
| 46 |
'model_w_link': model_hyperlink(self.model_link, self.model),
|
| 47 |
ModelInfoColumn.precision.name: self.precision.value.name,
|
| 48 |
ModelInfoColumn.model_type.name: self.model_type.value.name,
|
|
|
|
| 42 |
def to_dict(self):
|
| 43 |
"""Converts the model info to a dict compatible with our dataframe display"""
|
| 44 |
data_dict = {
|
| 45 |
+
ModelInfoColumn.model.name: self.model,
|
| 46 |
'model_w_link': model_hyperlink(self.model_link, self.model),
|
| 47 |
ModelInfoColumn.precision.name: self.precision.value.name,
|
| 48 |
ModelInfoColumn.model_type.name: self.model_type.value.name,
|