Spaces:
Running on Zero
Running on Zero
Upload folder using huggingface_hub
Browse files
app.py
CHANGED
|
@@ -399,7 +399,7 @@ with gr.Blocks(title="MLOL — MultiDomain LLM Optimisation Lab", css=CSS) as de
|
|
| 399 |
prep_btn = gr.Button("1️⃣ Prepare dataset", variant="primary")
|
| 400 |
with gr.Column(scale=2):
|
| 401 |
prep_md = gr.Markdown()
|
| 402 |
-
run_dd = gr.Dropdown(choices=run_choices(), label="Active experiment", interactive=True)
|
| 403 |
run_refresh = gr.Button("↻ refresh experiments", size="sm")
|
| 404 |
run_refresh.click(lambda: gr.update(choices=run_choices()), None, run_dd)
|
| 405 |
with gr.Accordion("Training configuration", open=True):
|
|
@@ -481,7 +481,7 @@ with gr.Blocks(title="MLOL — MultiDomain LLM Optimisation Lab", css=CSS) as de
|
|
| 481 |
gr.Markdown("Baseline vs fine-tuned with **identical items and seeds**, bootstrap CIs, and "
|
| 482 |
"paired permutation significance (α=0.05). Sampled evaluation — the certificate "
|
| 483 |
"always discloses n, seed, and that the full benchmark was not executed.")
|
| 484 |
-
ev_run = gr.Dropdown(choices=run_choices(), label="Experiment")
|
| 485 |
gr.Button("↻ refresh", size="sm").click(lambda: gr.update(choices=run_choices()), None, ev_run)
|
| 486 |
ev_view = gr.Button("Show stored evaluations")
|
| 487 |
ev_md = gr.Markdown()
|
|
@@ -509,7 +509,7 @@ with gr.Blocks(title="MLOL — MultiDomain LLM Optimisation Lab", css=CSS) as de
|
|
| 509 |
ev_view.click(show_evals, ev_run, ev_md)
|
| 510 |
|
| 511 |
with gr.Tab("📄 Reports"):
|
| 512 |
-
rp_run = gr.Dropdown(choices=run_choices(), label="Experiment")
|
| 513 |
gr.Button("↻ refresh", size="sm").click(lambda: gr.update(choices=run_choices()), None, rp_run)
|
| 514 |
rp_btn = gr.Button("Generate optimisation report + certificate", variant="primary")
|
| 515 |
rp_md = gr.Markdown()
|
|
@@ -520,8 +520,8 @@ with gr.Blocks(title="MLOL — MultiDomain LLM Optimisation Lab", css=CSS) as de
|
|
| 520 |
rp_btn.click(generate_reports, rp_run, [rp_md, rp_pdf, rp_csv, rp_json])
|
| 521 |
gr.Markdown("### Research dashboard — compare any two experiments")
|
| 522 |
with gr.Row():
|
| 523 |
-
cmp_a = gr.Dropdown(choices=run_choices(), label="Experiment A")
|
| 524 |
-
cmp_b = gr.Dropdown(choices=run_choices(), label="Experiment B")
|
| 525 |
cmp_btn = gr.Button("Compare")
|
| 526 |
cmp2_tbl = gr.Dataframe(interactive=False)
|
| 527 |
|
|
@@ -586,7 +586,7 @@ with gr.Blocks(title="MLOL — MultiDomain LLM Optimisation Lab", css=CSS) as de
|
|
| 586 |
if p.id == CFG.default_provider), None),
|
| 587 |
label="LLM provider")
|
| 588 |
as_key = gr.Textbox(label="API key (only for Claude/GPT; never stored)", type="password")
|
| 589 |
-
as_run = gr.Dropdown(choices=run_choices(), label="Experiment context (optional)")
|
| 590 |
gr.Button("↻", size="sm").click(lambda: gr.update(choices=run_choices()), None, as_run)
|
| 591 |
as_chat = gr.Chatbot(type="messages", height=280, label="Assistant")
|
| 592 |
as_msg = gr.Textbox(placeholder="Why did my model perform worse?", label="Ask")
|
|
|
|
| 399 |
prep_btn = gr.Button("1️⃣ Prepare dataset", variant="primary")
|
| 400 |
with gr.Column(scale=2):
|
| 401 |
prep_md = gr.Markdown()
|
| 402 |
+
run_dd = gr.Dropdown(choices=run_choices(), label="Active experiment", interactive=True, allow_custom_value=True)
|
| 403 |
run_refresh = gr.Button("↻ refresh experiments", size="sm")
|
| 404 |
run_refresh.click(lambda: gr.update(choices=run_choices()), None, run_dd)
|
| 405 |
with gr.Accordion("Training configuration", open=True):
|
|
|
|
| 481 |
gr.Markdown("Baseline vs fine-tuned with **identical items and seeds**, bootstrap CIs, and "
|
| 482 |
"paired permutation significance (α=0.05). Sampled evaluation — the certificate "
|
| 483 |
"always discloses n, seed, and that the full benchmark was not executed.")
|
| 484 |
+
ev_run = gr.Dropdown(choices=run_choices(), label="Experiment", allow_custom_value=True)
|
| 485 |
gr.Button("↻ refresh", size="sm").click(lambda: gr.update(choices=run_choices()), None, ev_run)
|
| 486 |
ev_view = gr.Button("Show stored evaluations")
|
| 487 |
ev_md = gr.Markdown()
|
|
|
|
| 509 |
ev_view.click(show_evals, ev_run, ev_md)
|
| 510 |
|
| 511 |
with gr.Tab("📄 Reports"):
|
| 512 |
+
rp_run = gr.Dropdown(choices=run_choices(), label="Experiment", allow_custom_value=True)
|
| 513 |
gr.Button("↻ refresh", size="sm").click(lambda: gr.update(choices=run_choices()), None, rp_run)
|
| 514 |
rp_btn = gr.Button("Generate optimisation report + certificate", variant="primary")
|
| 515 |
rp_md = gr.Markdown()
|
|
|
|
| 520 |
rp_btn.click(generate_reports, rp_run, [rp_md, rp_pdf, rp_csv, rp_json])
|
| 521 |
gr.Markdown("### Research dashboard — compare any two experiments")
|
| 522 |
with gr.Row():
|
| 523 |
+
cmp_a = gr.Dropdown(choices=run_choices(), label="Experiment A", allow_custom_value=True)
|
| 524 |
+
cmp_b = gr.Dropdown(choices=run_choices(), label="Experiment B", allow_custom_value=True)
|
| 525 |
cmp_btn = gr.Button("Compare")
|
| 526 |
cmp2_tbl = gr.Dataframe(interactive=False)
|
| 527 |
|
|
|
|
| 586 |
if p.id == CFG.default_provider), None),
|
| 587 |
label="LLM provider")
|
| 588 |
as_key = gr.Textbox(label="API key (only for Claude/GPT; never stored)", type="password")
|
| 589 |
+
as_run = gr.Dropdown(choices=run_choices(), label="Experiment context (optional)", allow_custom_value=True)
|
| 590 |
gr.Button("↻", size="sm").click(lambda: gr.update(choices=run_choices()), None, as_run)
|
| 591 |
as_chat = gr.Chatbot(type="messages", height=280, label="Assistant")
|
| 592 |
as_msg = gr.Textbox(placeholder="Why did my model perform worse?", label="Ask")
|