Update app.py
Browse files
app.py
CHANGED
|
@@ -70,7 +70,7 @@ eval_dataframe_test = get_dataframe_from_results(eval_results=eval_results, spli
|
|
| 70 |
# Gold answers
|
| 71 |
gold_results = {}
|
| 72 |
gold_dataset = load_dataset(INTERNAL_DATA_DATASET, "test", token=TOKEN, trust_remote_code=True)
|
| 73 |
-
gold_results = {"test": {row["
|
| 74 |
|
| 75 |
|
| 76 |
def restart_space():
|
|
@@ -153,19 +153,20 @@ def add_new_eval(
|
|
| 153 |
with open(f"scored/{organisation}_{model}.jsonl", "w") as scored_file:
|
| 154 |
with open(file_path, 'r') as f:
|
| 155 |
for ix, line in enumerate(f):
|
| 156 |
-
try:
|
| 157 |
-
|
| 158 |
-
except Exception:
|
| 159 |
-
|
| 160 |
-
if "model_answer" not in task:
|
| 161 |
-
|
| 162 |
-
answer = task["model_answer"]
|
| 163 |
-
task_id = task["task_id"]
|
| 164 |
-
|
| 165 |
-
if task_id not in gold_results[val_or_test]:
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
score = question_scorer(answer, gold_results[val_or_test][task_id]["Final answer"])
|
|
|
|
| 169 |
|
| 170 |
scored_file.write(
|
| 171 |
json.dumps({
|
|
|
|
| 70 |
# Gold answers
|
| 71 |
gold_results = {}
|
| 72 |
gold_dataset = load_dataset(INTERNAL_DATA_DATASET, "test", token=TOKEN, trust_remote_code=True)
|
| 73 |
+
gold_results = {"test": {row["cqs"]: row for row in gold_dataset["test"]}}
|
| 74 |
|
| 75 |
|
| 76 |
def restart_space():
|
|
|
|
| 153 |
with open(f"scored/{organisation}_{model}.jsonl", "w") as scored_file:
|
| 154 |
with open(file_path, 'r') as f:
|
| 155 |
for ix, line in enumerate(f):
|
| 156 |
+
#try:
|
| 157 |
+
# task = json.loads(line)
|
| 158 |
+
#except Exception:
|
| 159 |
+
# return format_error(f"Line {ix} is incorrectly formatted. Please fix it and resubmit your file.")
|
| 160 |
+
#if "model_answer" not in task:
|
| 161 |
+
# return format_error(f"Line {ix} missing 'model_answer'.")
|
| 162 |
+
#answer = task["model_answer"]
|
| 163 |
+
#task_id = task["task_id"]
|
| 164 |
+
|
| 165 |
+
#if task_id not in gold_results[val_or_test]:
|
| 166 |
+
# return format_error(f"{task_id} not found in gold set.")
|
| 167 |
+
|
| 168 |
+
#score = question_scorer(answer, gold_results[val_or_test][task_id]["Final answer"])
|
| 169 |
+
score = 1
|
| 170 |
|
| 171 |
scored_file.write(
|
| 172 |
json.dumps({
|