Update app.py (#18)

- Update app.py (ee7d6f710b5f37034acb2c2c5659c203f7e2899f)

app.py

@@ -6,6 +6,9 @@ import urllib
 
 usernames = {}
 
+filepath = hf_hub_download(repo_id="bigcode/the-stack-username-to-repo", filename="username_to_repo.json.gz", repo_type="dataset", revision="v2.0.1")
+with gzip.open(filepath, 'r') as f:
+    usernames["v2.0.1"] = json.loads(f.read().decode('utf-8'))
 
 filepath = hf_hub_download(repo_id="bigcode/the-stack-username-to-repo", filename="username_to_repo.json.gz", repo_type="dataset", revision="v2.0")
 with gzip.open(filepath, 'r') as f:
@@ -38,6 +41,11 @@ This tool lets you check if a repository under a given username is part of The S
 
 **Data source**:\
 <img src="https://annex.softwareheritage.org/public/logo/software-heritage-logo-title.2048px.png" alt="Logo" style="height: 3em; vertical-align: middle;" />
+
+**Model training**:\
+- StarCoder1 was trained on repos listed in `v1.2`.
+- StarCoder2 was trained on repos listed in `v2.0.1`.
+
 """
 opt_out_text_template = """\
 ### Opt-out
@@ -92,7 +100,7 @@ with gr.Blocks() as demo:
         _, colum_2, _ = gr.Column(scale=1), gr.Column(scale=6), gr.Column(scale=1)
         with colum_2:
             gr.Markdown(text)
-            version = gr.Dropdown(["v2.0", "v1.2", "v1.1", "v1.0"], label="The Stack version:", value="v2.0")
+            version = gr.Dropdown(["v2.0.1", "v2.0", "v1.2", "v1.1", "v1.0"], label="The Stack version:", value="v2.0.1")
             username = gr.Text("", label="Your GitHub username:")
             check_button = gr.Button("Check!")