sebastien
/
Copies
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Miscs fixes

master
Sébastien Miquel 2026-03-30 14:47:25 +02:00
parent f16c0273bd
commit bc59b2aeea
2 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
annotating_by_label.py
View File

@ -108,7 +108,7 @@ def main():
all_labels = os.path.join(root_dir, "labels") all_labels = os.path.join(root_dir, "labels")
if not os.path.exists(label_groups): if not os.path.exists(label_groups):
print(f"Warning: Labels file '{label_groups}' not found, making it out of '{all_labels}' (do check it…)") print(f"Warning: Labels file '{label_groups}' not found, making it out of '{all_labels}'")
if not os.path.exists(all_labels): if not os.path.exists(all_labels):
print(f"Error: {all_labels} not found.") print(f"Error: {all_labels} not found.")
sys.exit(1) sys.exit(1)

10
correction.py
View File

@ -184,7 +184,7 @@ def tprint(*args, **kwargs):
thread_logs[tid] = [] thread_logs[tid] = []
thread_logs[tid].append(msg) thread_logs[tid].append(msg)
Optional: Keep printing to console but prefix with thread name # Optional: Keep printing to console but prefix with thread name
print(f"[{tid}] {msg}", **kwargs) print(f"[{tid}] {msg}", **kwargs)
def flush_thread_log(tid=None): def flush_thread_log(tid=None):
@ -192,7 +192,7 @@ def flush_thread_log(tid=None):
tid = tid or threading.current_thread().name tid = tid or threading.current_thread().name
with log_lock: with log_lock:
if thread_logs.get(tid): if thread_logs.get(tid):
with open("correction_log", "a", encoding="utf-8") as f: with open(Path(INPUT_DIR) / "correction_log", "a", encoding="utf-8") as f:
f.write(f"--- Task Log [{tid}] ---\n") f.write(f"--- Task Log [{tid}] ---\n")
f.write("\n".join(thread_logs[tid]) + "\n\n") f.write("\n".join(thread_logs[tid]) + "\n\n")
thread_logs[tid].clear() thread_logs[tid].clear()
@ -313,7 +313,7 @@ def call_gemini_with_retries(model_id, contents, config,
# Immediately fallback to Flash without waiting if it's a Pro quota error # Immediately fallback to Flash without waiting if it's a Pro quota error
if is_quota_error and model_id == MODEL_ID_pro and fallback_model_id: if is_quota_error and model_id == MODEL_ID_pro and fallback_model_id:
tprint(f"\tGemini Pro quota hit ({e}). Falling back to Flash permanently...") tprint(f"\tGemini Pro quota hit ({e}). \n\n\tFalling back to Flash permanently...")
model_id = fallback_model_id model_id = fallback_model_id
pro_quota_exhausted = True pro_quota_exhausted = True
continue # Retry immediately with Flash continue # Retry immediately with Flash
@ -591,7 +591,7 @@ def process_single_task(task_tuple):
yming, ymaxg, width_r = d_data[pid] yming, ymaxg, width_r = d_data[pid]
pdf_path = Path(INPUT_DIR) / f"Copie{pid}" / f"{label}.pdf" pdf_path = Path(INPUT_DIR) / f"Copie{pid}" / f"{label}.pdf"
if !can_spawn_tasks and res["error"] == "additional-answer": if (not can_spawn_tasks) and res["error"] == "additional-answer":
tprint("\tSwallowing an additional-answer from a subsequent task.") tprint("\tSwallowing an additional-answer from a subsequent task.")
res["error"]= "" res["error"]= ""
if res["error"] != "": if res["error"] != "":
@ -602,7 +602,7 @@ def process_single_task(task_tuple):
needs_correction = [] needs_correction = []
for (i,f) in enumerate(res["feedback"]): for (i,f) in enumerate(res["feedback"]):
b = f["box_2d"] b = f.get("box_2d")
if b: if b:
ymin, xmin, ymax, xmax = b ymin, xmin, ymax, xmax = b
ymin = ymin * total_height // 1000 ymin = ymin * total_height // 1000