import streamlit as st import pandas as pd from ast import literal_eval import altair as alt import matplotlib.pyplot as plt from utils import process_dataset, eval_tags, change_and_delta from language import process_for_lang, filter_multilinguality from pipelines import filter_pipeline_data def main(): # Pick revision at top supported_revisions = ["03_07_23", "26_06_23","19_06_23", "12_06_23", "05_06_23", "29_05_23", "22_05_23", "15_05_23", "08_05_23", "01_05_23", "24_04_23", "17_04_23", "10_04_23", "03_04_23", "27_03_23", "20_03_23", "13_03_23", "06_03_23", "27_02_23", "20_02_23", "13_02_23","06_02_23", "30_01_23", "24_01_23", "16_01_23", "10_01_23", "02_01_23", "19_12_22", "12_12_22", "05_12_22", "28_11_22", "22_11_22", "14_11_22", "07_11_22", "31_10_22", "24_10_22", "17_10_22", "10_10_22", "27_09_22"] col1, col2, col3 = st.columns(3) with col1: new = st.selectbox( 'Last revision', supported_revisions, index=0) with col2: base = st.selectbox( 'Old revision', supported_revisions, index=1) with col3: base_old = st.selectbox( 'Very old revision', supported_revisions, index=2) # Process dataset old_old_data = process_dataset(base_old) old_data = process_dataset(base) data = process_dataset(new) old_old_data["tags"] = old_old_data.apply(eval_tags, axis=1) old_data["tags"] = old_data.apply(eval_tags, axis=1) data["tags"] = data.apply(eval_tags, axis=1) # High level count of models and rate of change total_samples_old_old = old_old_data.shape[0] total_samples_old = old_data.shape[0] total_samples = data.shape[0] curr_change, delta = change_and_delta(total_samples_old_old, total_samples_old, total_samples) col1, col2 = st.columns(2) with col1: st.metric(label="Total public models", value=total_samples, delta=total_samples-total_samples_old) with col2: st.metric(label="Rate of change", value=curr_change, delta=delta) # Tabs don't work in Spaces st version tab1, tab2, tab3, tab4, tab5, tab6, tab7, tab8 = st.tabs(["Language", "License", "Pipeline", "Social Features", "Libraries", "Model Cards", "Super users", "Raw Data"]) with tab1: st.header("Languages info") filtered_data = data.copy() old_filtered_data = old_data.copy() old_old_filtered_data = old_old_data.copy() modality = st.selectbox( 'Modalities', ["All", "NLP", "Audio", "Multimodal"]) filtered_data, no_lang_count, total_langs, langs = process_for_lang(filtered_data, modality) old_filtered_data, no_lang_count_old, total_langs_old, langs_old = process_for_lang(old_filtered_data, modality) old_old_filtered_data, no_lang_count_old_old, total_langs_old_old, _ = process_for_lang(old_old_filtered_data, modality) v = filtered_data.shape[0]-no_lang_count v_old = old_filtered_data.shape[0]-no_lang_count_old v_old_old = old_old_filtered_data.shape[0]-no_lang_count_old_old col1, col2 = st.columns(2) with col1: st.metric(label="Language Specified", value=v, delta=int(v-v_old)) with col2: curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="Language Specified Rate of Change", value=curr_change, delta=delta) col1, col2 = st.columns(2) with col1: st.metric(label="No Language Specified", value=no_lang_count, delta=int(no_lang_count-no_lang_count_old)) with col2: curr_change, delta = change_and_delta(no_lang_count_old_old, no_lang_count_old, no_lang_count) st.metric(label="No Language Specified Rate of Change", value=curr_change, delta=delta) col1, col2 = st.columns(2) with col1: st.metric(label="Total Unique Languages", value=total_langs, delta=int(total_langs-total_langs_old)) with col2: curr_change, delta = change_and_delta(total_langs_old_old, total_langs_old, total_langs) st.metric(label="Total Unique Languages Rate of Change", value=curr_change, delta=delta) st.text(f"New languages {set(langs)-set(langs_old)}") st.text(f"Lost languages {set(langs_old)-set(langs)}") st.subheader("Count of languages per model repo") st.text("Some repos are for multiple languages, so the count is greater than 1") linguality = st.selectbox( 'All or just Multilingual', ["All", "Just Multilingual", "Three or more languages"]) models_with_langs = filter_multilinguality(filtered_data, linguality) models_with_langs_old = filter_multilinguality(old_filtered_data, linguality) df1 = models_with_langs['language_count'].value_counts() df1_old = models_with_langs_old['language_count'].value_counts() st.bar_chart(df1) st.subheader("Most frequent languages") linguality_2 = st.selectbox( 'All or filtered', ["All", "No English", "Remove top 10"]) models_with_langs = filtered_data[filtered_data["language_count"] > 0] langs = models_with_langs["languages"].explode() langs = langs[langs != {}] orig_d = langs.value_counts().rename_axis("language").to_frame('counts').reset_index() d = orig_d models_with_langs_old = old_filtered_data[old_filtered_data["language_count"] > 0] langs = models_with_langs_old["languages"].explode() langs = langs[langs != {}] orig_d_old = langs.value_counts().rename_axis("language").to_frame('counts').reset_index() if linguality_2 == "No English": d = orig_d.iloc[1:] elif linguality_2 == "Remove top 10": d = orig_d.iloc[10:] # Just keep top 25 to avoid vertical scroll d = d.iloc[:25] st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('language', sort=None) )) st.subheader("Raw Data") l = df1.rename_axis("lang_count").reset_index().rename(columns={"language_count": "r_c"}) l_old = df1_old.rename_axis("lang_count").reset_index().rename(columns={"language_count": "old_r_c"}) final_data = pd.merge( l, l_old, how="outer", on="lang_count" ) print(final_data.head(3)) final_data["diff"] = final_data["r_c"] - final_data["old_r_c"] st.dataframe(final_data) d = orig_d.astype(str) orig_d_old = orig_d_old.astype(str).rename(columns={"counts": "old_c"}) final_data = pd.merge( d, orig_d_old, how="outer", on="language" ) final_data['counts'] = final_data['counts'].fillna(0).astype(int) final_data['old_c'] = final_data['old_c'].fillna(0).astype(int) final_data["diff"] = final_data["counts"] - final_data["old_c"] final_data['language'] = final_data['language'].astype(str) st.dataframe(final_data) with tab2: st.header("License info") no_license_count = data["license"].isna().sum() no_license_count_old = old_data["license"].isna().sum() no_license_count_old_old = old_old_data["license"].isna().sum() col1, col2 = st.columns(2) with col1: v = total_samples-no_license_count v_old = total_samples_old-no_license_count_old st.metric(label="License Specified", value=v, delta=int(v-v_old)) with col2: v = total_samples-no_license_count v_old = total_samples_old-no_license_count_old v_old_old = total_samples_old-no_license_count_old_old curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="License Specified Rate of Change", value=curr_change, delta=delta) col1, col2 = st.columns(2) with col1: st.metric(label="No License Specified", value=no_license_count, delta=int(no_license_count-no_license_count_old)) with col2: curr_change, delta = change_and_delta(no_license_count_old_old, no_license_count_old, no_license_count) st.metric(label="No License Specified Rate of Change", value=curr_change, delta=delta) col1, col2 = st.columns(2) unique_licenses = len(data["license"].unique()) unique_licenses_old = len(old_data["license"].unique()) unique_licenses_old_old = len(old_old_data["license"].unique()) with col1: st.metric(label="Total Unique Licenses", value=unique_licenses, delta=int(unique_licenses-unique_licenses_old)) with col2: curr_change, delta = change_and_delta(unique_licenses_old_old, unique_licenses_old, unique_licenses) st.metric(label="Total Unique Licenses Rate of Change", value=curr_change, delta=delta) st.text(f"New licenses {set(data['license'].unique())-set(old_data['license'].unique())}") st.text(f"Old licenses {set(old_data['license'].unique())-set(data['license'].unique())}") st.subheader("Distribution of licenses per model repo") license_filter = st.selectbox( 'All or filtered', ["All", "No Apache 2.0", "Remove top 10"]) filter = 0 if license_filter == "All": filter = 0 elif license_filter == "No Apache 2.0": filter = 1 else: filter = 2 d = data["license"].value_counts().rename_axis("license").to_frame('counts').reset_index() if filter == 1: d = d.iloc[1:] elif filter == 2: d = d.iloc[10:] # Just keep top 25 to avoid vertical scroll d = d.iloc[:25] st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('license', sort=None) )) st.text("There are some edge cases, as old repos using lists of licenses.") st.subheader("Raw Data") d = data["license"].value_counts().rename_axis("license").to_frame('counts').reset_index() d_old = old_data["license"].value_counts().rename_axis("license").to_frame('counts').reset_index().rename(columns={"counts": "old_c"}) final_data = pd.merge( d, d_old, how="outer", on="license" ) final_data["diff"] = final_data["counts"] - final_data["old_c"] st.dataframe(final_data) with tab3: st.header("Pipeline info") tags = data["tags"].explode() tags = tags[tags.notna()].value_counts().rename_axis("tag").to_frame('counts').reset_index() s = tags["tag"] s = s[s.apply(type) == str] unique_tags = len(s.unique()) tags_old = old_data["tags"].explode() tags_old = tags_old[tags_old.notna()].value_counts().rename_axis("tag").to_frame('counts').reset_index() s_o = tags_old["tag"] s_o = s_o[s_o.apply(type) == str] unique_tags_old = len(s_o.unique()) tags_old_old = old_old_data["tags"].explode() tags_old_old = tags_old_old[tags_old_old.notna()].value_counts().rename_axis("tag").to_frame('counts').reset_index() s_old_old = tags_old_old["tag"] s_old_old = s_old_old[s_old_old.apply(type) == str] unique_tags_old_old = len(s_old_old.unique()) no_pipeline_count = data["pipeline"].isna().sum() no_pipeline_count_old = old_data["pipeline"].isna().sum() no_pipeline_count_old_old = old_old_data["pipeline"].isna().sum() col1, col2 = st.columns(2) v = total_samples-no_pipeline_count v_old = total_samples_old-no_pipeline_count_old v_old_old = total_samples_old_old-no_pipeline_count_old_old with col1: st.metric(label="# models that have any pipeline", value=v, delta=int(v-v_old)) with col2: curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="# models rate of change", value=curr_change, delta=delta) col1, col2 = st.columns(2) with col1: st.metric(label="No pipeline Specified", value=no_pipeline_count, delta=int(no_pipeline_count-no_pipeline_count_old)) with col2: curr_change, delta = change_and_delta(no_pipeline_count_old_old, no_pipeline_count_old, no_pipeline_count) st.metric(label="No pipeline Specified rate of change", value=curr_change, delta=delta) col1, col2 = st.columns(2) with col1: st.metric(label="Total Unique Tags", value=unique_tags, delta=int(unique_tags-unique_tags_old)) with col2: curr_change, delta = change_and_delta(unique_tags_old_old, unique_tags_old, unique_tags) st.metric(label="Total Unique Tags", value=curr_change, delta=delta) modality_filter = st.selectbox( 'Modalities', ["All", "NLP", "CV", "Audio", "RL", "Multimodal", "Tabular"]) st.subheader("High-level metrics") col1, col2, col3 = st.columns(3) with col1: p = st.selectbox( 'What pipeline do you want to see?', ["all", *data["pipeline"].unique()] ) with col2: l = st.selectbox( 'What library do you want to see?', ["all", "not transformers", *data["library"].unique()] ) with col3: f = st.selectbox( 'What trf framework support?', ["all", "pytorch", "tensorflow", "jax"] ) col1, col2 = st.columns(2) with col1: filt = st.multiselect( label="Tags (All by default)", options=s.unique(), default=None) with col2: o = st.selectbox( label="Operation (for tags)", options=["Any", "All", "None"] ) filtered_data, tags = filter_pipeline_data(data, modality_filter, p, l, f, filt, o) filtered_data_old, old_tags = filter_pipeline_data(old_data, modality_filter, p, l, f, filt, o) filtered_data_old_old, old_old_tags = filter_pipeline_data(old_old_data, modality_filter, p, l, f, filt, o) st.subheader("Pipeline breakdown") d = filtered_data["pipeline"].value_counts().rename_axis("pipeline").to_frame('counts').reset_index() columns_of_interest = ["downloads_30d", "likes", "pytorch", "tensorflow", "jax"] grouped_data = filtered_data.groupby("pipeline").sum()[columns_of_interest] final_data = pd.merge( d, grouped_data, how="outer", on="pipeline" ) d_old = filtered_data_old["pipeline"].value_counts().rename_axis("pipeline").to_frame('counts').reset_index() grouped_data_old = filtered_data_old.groupby("pipeline").sum()[columns_of_interest] final_data_old = pd.merge( d_old, grouped_data_old, how="outer", on="pipeline" ) d_old = filtered_data_old_old["pipeline"].value_counts().rename_axis("pipeline").to_frame('counts').reset_index() grouped_data_old_old = filtered_data_old_old.groupby("pipeline").sum()[columns_of_interest] sums = grouped_data.sum() sums_old = grouped_data_old.sum() sums_old_old = grouped_data_old_old.sum() col1, col2, col3, col4 = st.columns(4) v = filtered_data.shape[0] v_old = filtered_data_old.shape[0] v_old_old = filtered_data_old_old.shape[0] with col1: st.metric(label="Total models", value=v, delta=int(v - v_old)) with col2: curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="Total models rate of change", value=curr_change, delta=delta) with col3: st.metric(label="Cumulative Downloads (30d)", value=sums["downloads_30d"], delta=int(sums["downloads_30d"] - sums_old["downloads_30d"])) with col4: print(sums_old_old["downloads_30d"], sums_old["downloads_30d"], sums["downloads_30d"]) curr_change, delta = change_and_delta(sums_old_old["downloads_30d"], sums_old["downloads_30d"], sums["downloads_30d"]) st.metric(label="Cumulative Downloads (30d) rate of change", value=curr_change, delta=delta) col1, col2, col3 = st.columns(3) with col1: st.metric(label="Total unique pipelines", value=len(filtered_data["pipeline"].unique())) with col2: st.metric(label="Cumulative likes", value=sums["likes"], delta=int(sums["likes"] - sums_old["likes"])) with col3: curr_change, delta = change_and_delta(sums_old_old["likes"], sums_old["likes"], sums["likes"]) st.metric(label="Cumulative Likes rate of change", value=curr_change, delta=delta) col1, col2, col3 = st.columns(3) with col1: st.metric(label="Total in PT", value=sums["pytorch"], delta=int(sums["pytorch"] - sums_old["pytorch"])) with col2: st.metric(label="Total in TF", value=sums["tensorflow"], delta=int(sums["tensorflow"] - sums_old["tensorflow"])) with col3: st.metric(label="Total in JAX", value=sums["jax"], delta=int(sums["jax"] - sums_old["jax"])) col1, col2 = st.columns(2) with col1: st.metric(label="Total unique libraries", value=len(filtered_data["library"].unique())) with col2: st.metric(label="Total unique modality", value=len(filtered_data["modality"].unique())) col1, col2 = st.columns(2) with col1: st.metric(label="Total transformers models", value=len(filtered_data[filtered_data["library"] == "transformers"])) with col2: st.metric(label="Total non transformers models", value=len(filtered_data[filtered_data["library"] != "transformers"])) st.metric(label="Unique Tags", value=len(tags), delta=int(len(tags) - len(old_tags))) st.text(f"New tags {set(tags)-set(old_tags)}") st.text(f"Lost tags {set(old_tags)-set(tags)}") st.subheader("Pipeline breakdown by modality") col1, col2 = st.columns(2) with col1: st.metric(label="Total CV models", value=len(filtered_data[filtered_data["modality"] == "cv"])) with col2: st.metric(label="Total NLP models", value=len(filtered_data[filtered_data["modality"] == "nlp"])) col1, col2 = st.columns(2) with col1: st.metric(label="Total Audio models", value=len(filtered_data[filtered_data["modality"] == "audio"])) with col2: st.metric(label="Total RL models", value=len(filtered_data[filtered_data["modality"] == "rl"])) col1, col2 = st.columns(2) with col1: st.metric(label="Total Tabular models", value=len(filtered_data[filtered_data["modality"] == "tabular"])) with col2: st.metric(label="Total Multimodal models", value=len(filtered_data[filtered_data["modality"] == "multimodal"])) st.subheader("Count of models per pipeline") st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('pipeline', sort=None) )) st.subheader("Aggregated data") st.dataframe(final_data) st.subheader("Most common model types (specific to transformers)") d = filtered_data["model_type"].value_counts().rename_axis("model_type").to_frame('counts').reset_index() d = d.iloc[:15] st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('model_type', sort=None) )) st.subheader("Most common library types (Learn more in library tab)") d = filtered_data["library"].value_counts().rename_axis("library").to_frame('counts').reset_index().head(15) st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('library', sort=None) )) st.subheader("Tags by count") tags = filtered_data["tags"].explode() tags = tags[tags.notna()].value_counts().rename_axis("tag").to_frame('counts').reset_index() st.write(alt.Chart(tags.head(30)).mark_bar().encode( x='counts', y=alt.X('tag', sort=None) )) st.subheader("Raw Data") columns_of_interest = [ "repo_id", "author", "model_type", "files_per_repo", "library", "downloads_30d", "likes", "pytorch", "tensorflow", "jax"] raw_data = filtered_data[columns_of_interest] st.dataframe(raw_data) # todo : add activity metric with tab4: st.header("Social Features") columns_of_interest = ["prs_count", "prs_open", "prs_merged", "prs_closed", "discussions_count", "discussions_open", "discussions_closed"] sums = data[columns_of_interest].sum() sums_old = old_data[columns_of_interest].sum() sums_old_old = old_old_data[columns_of_interest].sum() col1, col2, col3, col4 = st.columns(4) with col1: st.metric(label="Total PRs", value=sums["prs_count"],delta=int(sums["prs_count"] - sums_old["prs_count"])) with col2: st.metric(label="PRs opened", value=sums["prs_open"], delta=int(sums["prs_open"] - sums_old["prs_open"])) with col3: st.metric(label="PRs merged", value=sums["prs_merged"], delta=int(sums["prs_merged"] - sums_old["prs_merged"])) with col4: st.metric(label="PRs closed", value=sums["prs_closed"], delta=int(sums["prs_closed"] - sums_old["prs_closed"])) col1, col2, col3, col4 = st.columns(4) with col1: curr_change, delta = change_and_delta(sums_old_old["prs_count"], sums_old["prs_count"], sums["prs_count"]) st.metric(label="Total PRs change", value=curr_change,delta=delta) with col2: curr_change, delta = change_and_delta(sums_old_old["prs_open"], sums_old["prs_open"], sums["prs_open"]) st.metric(label="PRs opened change", value=curr_change,delta=delta) with col3: curr_change, delta = change_and_delta(sums_old_old["prs_merged"], sums_old["prs_merged"], sums["prs_merged"]) st.metric(label="PRs merged change", value=curr_change,delta=delta) with col4: curr_change, delta = change_and_delta(sums_old_old["prs_closed"], sums_old["prs_closed"], sums["prs_closed"]) st.metric(label="PRs closed change", value=curr_change,delta=delta) col1, col2, col3 = st.columns(3) with col1: st.metric(label="Total discussions", value=sums["discussions_count"], delta=int(sums["discussions_count"] - sums_old["discussions_count"])) with col2: st.metric(label="Discussions open", value=sums["discussions_open"], delta=int(sums["discussions_open"] - sums_old["discussions_open"])) with col3: st.metric(label="Discussions closed", value=sums["discussions_closed"], delta=int(sums["discussions_closed"] - sums_old["discussions_closed"])) col1, col2, col3 = st.columns(3) with col1: curr_change, delta = change_and_delta(sums_old_old["discussions_count"], sums_old["discussions_count"], sums["discussions_count"]) st.metric(label="Total discussions change", value=curr_change,delta=delta) with col2: curr_change, delta = change_and_delta(sums_old_old["discussions_open"], sums_old["discussions_open"], sums["discussions_open"]) st.metric(label="Discussions open change", value=curr_change,delta=delta) with col3: curr_change, delta = change_and_delta(sums_old_old["discussions_closed"], sums_old["discussions_closed"], sums["discussions_closed"]) st.metric(label="Discussions closed change", value=curr_change,delta=delta) likes = [] for r in supported_revisions: likes.append(process_dataset(r)["likes"].sum()) source = pd.DataFrame({ 'revision': supported_revisions[::-1], 'likes': likes[::-1], }) st.subheader("Total likes") st.write(alt.Chart(source).mark_bar().encode( x=alt.X('revision', sort=alt.EncodingSortField(field="revision", op="count", order='ascending')), y='likes' )) st.subheader("Likes Rate of Change") diffs = source["likes"].pct_change() source = pd.DataFrame({ 'revision': supported_revisions[::-1][1:], 'likes_change': diffs[1:], }) print(source[["revision", "likes_change"]]) st.write(alt.Chart(source).mark_bar().encode( x=alt.X('revision', sort=alt.EncodingSortField(field="revision", op="count", order='ascending')), y='likes_change' )) st.subheader("Raw Data") filtered_data = data[["repo_id", "prs_count", "prs_open", "prs_merged", "prs_closed", "discussions_count", "discussions_open", "discussions_closed"]].sort_values("prs_count", ascending=False).reset_index(drop=True) st.dataframe(filtered_data) with tab5: st.header("Library info") no_library_count = data["library"].isna().sum() no_library_count_old = old_data["library"].isna().sum() no_library_count_old_old = old_old_data["library"].isna().sum() col1, col2, col3 = st.columns(3) with col1: v = total_samples-no_library_count v_old = total_samples_old-no_library_count_old st.metric(label="# models that have any library", value=v, delta=int(v-v_old)) with col2: st.metric(label="No library Specified", value=no_library_count, delta=int(no_library_count-no_library_count_old)) with col3: v = len(data["library"].unique()) v_old = len(old_data["library"].unique()) st.metric(label="Total Unique library", value=v, delta=int(v-v_old)) col1, col2, col3 = st.columns(3) with col1: v = total_samples-no_library_count v_old = total_samples_old-no_library_count_old v_old_old = total_samples_old_old-no_library_count_old_old curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="# models that have any library change", value=curr_change, delta=delta) with col2: curr_change, delta = change_and_delta(no_library_count_old_old, no_library_count_old, no_library_count) st.metric(label="No library Specified Change", value=curr_change, delta=delta) with col3: v = len(data["library"].unique()) v_old = len(old_data["library"].unique()) v_old_old = len(old_old_data["library"].unique()) curr_change, delta = change_and_delta(v_old_old, v_old, v) st.metric(label="Total Unique library", value=curr_change, delta=delta) st.subheader("High-level metrics") filtered_data = data[data['library'].notna()] filtered_data_old = old_data[old_data['library'].notna()] col1, col2 = st.columns(2) with col1: lib = st.selectbox( 'What library do you want to see? ', ["all", "not transformers", *filtered_data["library"].unique()] ) with col2: pip = st.selectbox( 'What pipeline do you want to see? ', ["all", *filtered_data["pipeline"].unique()] ) if pip != "all" : filtered_data = filtered_data[filtered_data["pipeline"] == pip] filtered_data_old = filtered_data_old[filtered_data_old["pipeline"] == pip] if lib != "all" and lib != "not transformers": filtered_data = filtered_data[filtered_data["library"] == lib] filtered_data_old = filtered_data_old[filtered_data_old["library"] == lib] if lib == "not transformers": filtered_data = filtered_data[filtered_data["library"] != "transformers"] filtered_data_old = filtered_data_old[filtered_data_old["library"] != "transformers"] d = filtered_data["library"].value_counts().rename_axis("library").to_frame('counts').reset_index() grouped_data = filtered_data.groupby("library").sum()[["downloads_30d", "likes"]] final_data = pd.merge( d, grouped_data, how="outer", on="library" ) sums = grouped_data.sum() d_old = filtered_data_old["library"].value_counts().rename_axis("library").to_frame('counts').reset_index() grouped_data_old = filtered_data_old.groupby("library").sum()[["downloads_30d", "likes"]] final_data_old = pd.merge( d_old, grouped_data_old, how="outer", on="library" ).add_suffix('_old') final_data_old = final_data_old.rename(index=str, columns={"library_old": "library"}) sums_old = grouped_data_old.sum() col1, col2, col3 = st.columns(3) with col1: v = filtered_data.shape[0] v_old = filtered_data_old.shape[0] st.metric(label="Total models", value=v, delta=int(v-v_old)) with col2: st.metric(label="Cumulative Downloads (30d)", value=sums["downloads_30d"], delta=int(sums["downloads_30d"]-sums_old["downloads_30d"])) with col3: st.metric(label="Cumulative likes", value=sums["likes"], delta=int(sums["likes"]-sums_old["likes"])) st.subheader("Most common library types (Learn more in library tab)") d = filtered_data["library"].value_counts().rename_axis("library").to_frame('counts').reset_index().head(15) st.write(alt.Chart(d).mark_bar().encode( x='counts', y=alt.X('library', sort=None) )) st.subheader("Aggregated Data") final_data = pd.merge( final_data, final_data_old, how="outer", on="library" ) final_data["counts_diff"] = final_data["counts"] - final_data["counts_old"] final_data["downloads_diff"] = final_data["downloads_30d"] - final_data["downloads_30d_old"] final_data["likes_diff"] = final_data["likes"] - final_data["likes_old"] st.dataframe(final_data) st.subheader("Raw Data") columns_of_interest = ["repo_id", "author", "files_per_repo", "library", "downloads_30d", "likes"] filtered_data = filtered_data[columns_of_interest] st.dataframe(filtered_data) with tab6: st.header("Model cards") columns_of_interest = ["has_model_index", "has_metadata", "has_text", "text_length"] rows = data.shape[0] rows_old = old_data.shape[0] rows_old_old = old_old_data.shape[0] cond = data["has_model_index"] | data["has_text"] with_model_card = data[cond] c_model_card = with_model_card.shape[0] cond = old_data["has_model_index"] | old_data["has_text"] with_model_card_old = old_data[cond] c_model_card_old = with_model_card_old.shape[0] cond = old_old_data["has_model_index"] | old_old_data["has_text"] with_model_card_old_old = old_old_data[cond] c_model_card_old_old = with_model_card_old_old.shape[0] st.subheader("High-level metrics") col1, col2, col3, col4 = st.columns(4) with col1: st.metric(label="# with model card file", value=c_model_card, delta=int(c_model_card-c_model_card_old)) with col2: curr_change, delta = change_and_delta(c_model_card_old_old, c_model_card_old, c_model_card) st.metric(label="# with model card file change", value=curr_change, delta=delta) with col3: st.metric(label="# without model card file", value=rows-c_model_card, delta=int((rows-c_model_card)-(rows_old-c_model_card_old))) with col4: curr_change, delta = change_and_delta(rows_old_old-c_model_card_old_old, rows_old-c_model_card_old, rows-c_model_card) st.metric(label="# without model card file change", value=curr_change, delta=delta) with_index = data["has_model_index"].sum() with_index_old = old_data["has_model_index"].sum() with_index_old_old = old_old_data["has_model_index"].sum() with col1: st.metric(label="# with model index", value=with_index, delta=int(with_index-with_index_old)) with col2: curr_change, delta = change_and_delta(with_index_old_old, with_index_old, with_index) st.metric(label="# with model index change", value=curr_change, delta=delta) with col3: st.metric(label="# without model index", value=rows-with_index, delta=int((rows-with_index)-(rows_old-with_index_old))) with col4: curr_change, delta = change_and_delta(rows_old_old-with_index_old_old, rows_old-with_index_old, rows-with_index) st.metric(label="# without model index change", value=curr_change, delta=delta) with_text = data["has_text"] with_text_old = old_data["has_text"] with_text_old_old = old_old_data["has_text"] with_text_sum = with_text.sum() with_text_old_sum = with_text_old.sum() with_text_old_old_sum = with_text_old_old.sum() with col1: st.metric(label="# with model card text", value=with_text_sum, delta=int(with_text_sum-with_text_old_sum)) with col2: curr_change, delta = change_and_delta(with_text_old_old_sum, with_text_old_sum, with_text_sum) st.metric(label="# with model card text change", value=curr_change, delta=delta) with col3: st.metric(label="# without card text", value=rows-with_text_sum, delta=int((rows-with_text_sum)-(with_text_old_sum))) with col4: curr_change, delta = change_and_delta(rows_old_old-with_text_old_old_sum, rows_old-with_text_old_sum, rows-with_text_sum) st.metric(label="# without card text change", value=curr_change, delta=delta) st.subheader("Length (chars) of model card content") fig, _ = plt.subplots() _ = data["length_bins"].value_counts().plot.bar() st.metric(label="# average length of model card (chars)", value=data[with_text]["text_length"].mean()) st.pyplot(fig) st.subheader("Tags (Read more in Pipeline tab)") tags = data["tags"].explode() tags = tags[tags.notna()].value_counts().rename_axis("tag").to_frame('counts').reset_index() st.write(alt.Chart(tags.head(30)).mark_bar().encode( x='counts', y=alt.X('tag', sort=None) )) with tab7: st.header("Authors") st.text("This info corresponds to the repos owned by the authors") authors = data.groupby("author").sum().drop(["text_length", "Unnamed: 0"], axis=1).sort_values("downloads_30d", ascending=False) d = data["author"].value_counts().rename_axis("author").to_frame('counts').reset_index() final_data = pd.merge( d, authors, how="outer", on="author" ) st.dataframe(final_data) with tab8: st.header("Raw Data") d = data.astype(str) st.dataframe(d) if __name__ == '__main__': main()