Spaces:
Sleeping
Sleeping
Update current results
Browse files- app.py +1 -1
- evaluate.py +2 -2
- results.json +50 -1
app.py
CHANGED
|
@@ -58,7 +58,7 @@ with demo:
|
|
| 58 |
dropdown_1 = gr.Dropdown(choices = models, value = models[0])
|
| 59 |
button_1 = gr.Button("Submit")
|
| 60 |
with gr.Row():
|
| 61 |
-
chart = gr.LinePlot(df, "Step", "Loss", color="Test", x_lim = (0,
|
| 62 |
|
| 63 |
button_1.click(return_results, dropdown_1, chart)
|
| 64 |
|
|
|
|
| 58 |
dropdown_1 = gr.Dropdown(choices = models, value = models[0])
|
| 59 |
button_1 = gr.Button("Submit")
|
| 60 |
with gr.Row():
|
| 61 |
+
chart = gr.LinePlot(df, "Step", "Loss", color="Test", x_lim = (0, max(df['Step'])))
|
| 62 |
|
| 63 |
button_1.click(return_results, dropdown_1, chart)
|
| 64 |
|
evaluate.py
CHANGED
|
@@ -10,7 +10,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
| 10 |
device = "cuda"
|
| 11 |
test_indices_length = 1000
|
| 12 |
|
| 13 |
-
models = ["distributed/optimized-gpt2-250m", "distributed/optimized-gpt2-250m-v0.1.1", "distributed/gpt2-94m"]
|
| 14 |
|
| 15 |
if os.path.exists("results.json"):
|
| 16 |
with open('results.json', 'r') as file:
|
|
@@ -28,7 +28,7 @@ for model_name in models:
|
|
| 28 |
refs = list_repo_refs(model_name, repo_type="model")
|
| 29 |
global_epoch = max([int(tag.name) for tag in refs.tags]) if refs.tags else None
|
| 30 |
|
| 31 |
-
for epoch in range(0,global_epoch,
|
| 32 |
|
| 33 |
if str(epoch) in results[model_name]['main-net'].keys():
|
| 34 |
continue
|
|
|
|
| 10 |
device = "cuda"
|
| 11 |
test_indices_length = 1000
|
| 12 |
|
| 13 |
+
models = ["distributed/optimized-gpt2-250m", "distributed/optimized-gpt2-250m-v0.1.3", "distributed/optimized-gpt2-250m-v0.1.1", "distributed/gpt2-94m"]
|
| 14 |
|
| 15 |
if os.path.exists("results.json"):
|
| 16 |
with open('results.json', 'r') as file:
|
|
|
|
| 28 |
refs = list_repo_refs(model_name, repo_type="model")
|
| 29 |
global_epoch = max([int(tag.name) for tag in refs.tags]) if refs.tags else None
|
| 30 |
|
| 31 |
+
for epoch in range(0,global_epoch, 1):
|
| 32 |
|
| 33 |
if str(epoch) in results[model_name]['main-net'].keys():
|
| 34 |
continue
|
results.json
CHANGED
|
@@ -1,6 +1,55 @@
|
|
| 1 |
{
|
| 2 |
"distributed/optimized-gpt2-250m": {
|
| 3 |
-
"main-net": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
"baseline": {
|
| 5 |
"0": [
|
| 6 |
10.95511245727539
|
|
|
|
| 1 |
{
|
| 2 |
"distributed/optimized-gpt2-250m": {
|
| 3 |
+
"main-net": {
|
| 4 |
+
"0": [
|
| 5 |
+
10.996320237222296
|
| 6 |
+
],
|
| 7 |
+
"1": [
|
| 8 |
+
10.96618640299416
|
| 9 |
+
],
|
| 10 |
+
"2": [
|
| 11 |
+
10.90651673192917
|
| 12 |
+
],
|
| 13 |
+
"3": [
|
| 14 |
+
10.81940748757928
|
| 15 |
+
],
|
| 16 |
+
"4": [
|
| 17 |
+
10.712032748685024
|
| 18 |
+
],
|
| 19 |
+
"5": [
|
| 20 |
+
10.591520558323776
|
| 21 |
+
],
|
| 22 |
+
"6": [
|
| 23 |
+
10.446569502482404
|
| 24 |
+
],
|
| 25 |
+
"7": [
|
| 26 |
+
10.32113439958174
|
| 27 |
+
],
|
| 28 |
+
"8": [
|
| 29 |
+
10.1767857876407
|
| 30 |
+
],
|
| 31 |
+
"9": [
|
| 32 |
+
10.047991905456934
|
| 33 |
+
],
|
| 34 |
+
"10": [
|
| 35 |
+
9.923473688189903
|
| 36 |
+
],
|
| 37 |
+
"11": [
|
| 38 |
+
9.804602791635613
|
| 39 |
+
],
|
| 40 |
+
"12": [
|
| 41 |
+
9.697358704748607
|
| 42 |
+
],
|
| 43 |
+
"13": [
|
| 44 |
+
9.610764950228923
|
| 45 |
+
],
|
| 46 |
+
"14": [
|
| 47 |
+
9.533995848289777
|
| 48 |
+
],
|
| 49 |
+
"15": [
|
| 50 |
+
9.494339939102112
|
| 51 |
+
]
|
| 52 |
+
},
|
| 53 |
"baseline": {
|
| 54 |
"0": [
|
| 55 |
10.95511245727539
|