From: François Fleuret Date: Thu, 12 Sep 2024 17:14:47 +0000 (+0200) Subject: Update. X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=commitdiff_plain;h=68eb9d66255c19be3412cf4cb28534d5bc165697;p=culture.git Update. --- diff --git a/main.py b/main.py index f6f37d6..80e99fd 100755 --- a/main.py +++ b/main.py @@ -1228,6 +1228,8 @@ def thread_generate_ae_c_quizzes(models, nb, record, local_device=main_device): def save_c_quizzes_with_scores(models, c_quizzes, filename, solvable_only=False): l = [] + c_quizzes = c_quizzes.to(main_device) + with torch.autograd.no_grad(): if solvable_only: to_keep, _ = quiz_validation( @@ -1374,43 +1376,23 @@ def multithread_execution(fun, arguments): for t in threads: t.join() - if records == []: + if records[0] is None: return + else: return [ - torch.cat([x[k] for x in records], dim=0) for k in range(len(records[0])) + torch.cat([x[k].to("cpu") for x in records], dim=0) + for k in range(len(records[0])) ] # ----- test -# nb_gpus = len(gpus) -# nb_c_quizzes_to_generate = (args.nb_c_quizzes + nb_gpus - 1) // nb_gpus - -# c_quizzes, agreements = multithread_execution( -# generate_ae_c_quizzes, -# [(models, nb_c_quizzes_to_generate, gpu) for gpu in gpus], -# ) - # ranked_models = sorted(models, key=lambda m: float(m.test_accuracy)) # weakest_models = ranked_models[: len(gpus)] # n_epoch = 14 -# multithread_execution( -# one_ae_epoch, -# [ -# ( -# model, -# quiz_machine, -# n_epoch, -# None if c_quizzes is None else c_quizzes[agreements[:, model.id]], -# gpu, -# ) -# for model, gpu in zip(weakest_models, gpus) -# ], -# ) - ###################################################################### for n_epoch in range(current_epoch, args.nb_epochs): @@ -1459,35 +1441,40 @@ for n_epoch in range(current_epoch, args.nb_epochs): # -------------------------------------------------------------------- - records, threads = [], [] - - start_time = time.perf_counter() - - if len(gpus) > 1: - for gpu in gpus: - t = threading.Thread( - target=thread_generate_ae_c_quizzes, - daemon=True, - args=(models, nb_c_quizzes_to_generate, records, gpu), - ) - - # To get a different sequence between threads - log_string(f"dummy {torch.rand(1)}") - threads.append(t) - t.start() - - for t in threads: - t.join() - - else: - records.append( - generate_ae_c_quizzes(models, nb_c_quizzes_to_generate, gpus[0]) - ) - - time_c_quizzes = int(time.perf_counter() - start_time) + c_quizzes, agreements = multithread_execution( + generate_ae_c_quizzes, + [(models, nb_c_quizzes_to_generate, gpu) for gpu in gpus], + ) - c_quizzes = torch.cat([q.to(main_device) for q, _ in records], dim=0) - agreements = torch.cat([a.to(main_device) for _, a in records], dim=0) + ## records, threads = [], [] + ## + ## start_time = time.perf_counter() + ## + ## if len(gpus) > 1: + ## for gpu in gpus: + ## t = threading.Thread( + ## target=thread_generate_ae_c_quizzes, + ## daemon=True, + ## args=(models, nb_c_quizzes_to_generate, records, gpu), + ## ) + ## + ## # To get a different sequence between threads + ## log_string(f"dummy {torch.rand(1)}") + ## threads.append(t) + ## t.start() + ## + ## for t in threads: + ## t.join() + ## + ## else: + ## records.append( + ## generate_ae_c_quizzes(models, nb_c_quizzes_to_generate, gpus[0]) + ## ) + ## + ## time_c_quizzes = int(time.perf_counter() - start_time) + ## + ## c_quizzes = torch.cat([q.to(main_device) for q, _ in records], dim=0) + ## agreements = torch.cat([a.to(main_device) for _, a in records], dim=0) # -------------------------------------------------------------------- @@ -1520,40 +1507,54 @@ for n_epoch in range(current_epoch, args.nb_epochs): ranked_models = sorted(models, key=lambda m: float(m.test_accuracy)) weakest_models = ranked_models[: len(gpus)] - threads = [] - - start_time = time.perf_counter() - - if len(gpus) > 1: - for gpu, model in zip(gpus, weakest_models): - log_string(f"training model {model.id} (accuracy {model.test_accuracy})") - if c_quizzes is None: - c_quizzes_for_this_model = None - else: - c_quizzes_for_this_model = c_quizzes[agreements[:, model.id]] - - t = threading.Thread( - target=one_ae_epoch, - daemon=True, - args=(model, quiz_machine, n_epoch, c_quizzes_for_this_model, gpu), + multithread_execution( + one_ae_epoch, + [ + ( + model, + quiz_machine, + n_epoch, + None if c_quizzes is None else c_quizzes[agreements[:, model.id]], + gpu, ) + for model, gpu in zip(weakest_models, gpus) + ], + ) - threads.append(t) - - t.start() - - for t in threads: - t.join() - - else: - model = weakest_models[0] - log_string(f"training model {model.id} (accuracy {model.test_accuracy})") - if c_quizzes is None: - c_quizzes_for_this_model = None - else: - c_quizzes_for_this_model = c_quizzes[agreements[:, model.id]] - - one_ae_epoch(model, quiz_machine, n_epoch, c_quizzes_for_this_model, gpus[0]) + ## threads = [] + ## + ## start_time = time.perf_counter() + ## + ## if len(gpus) > 1: + ## for gpu, model in zip(gpus, weakest_models): + ## log_string(f"training model {model.id} (accuracy {model.test_accuracy})") + ## if c_quizzes is None: + ## c_quizzes_for_this_model = None + ## else: + ## c_quizzes_for_this_model = c_quizzes[agreements[:, model.id]] + ## + ## t = threading.Thread( + ## target=one_ae_epoch, + ## daemon=True, + ## args=(model, quiz_machine, n_epoch, c_quizzes_for_this_model, gpu), + ## ) + ## + ## threads.append(t) + ## + ## t.start() + ## + ## for t in threads: + ## t.join() + ## + ## else: + ## model = weakest_models[0] + ## log_string(f"training model {model.id} (accuracy {model.test_accuracy})") + ## if c_quizzes is None: + ## c_quizzes_for_this_model = None + ## else: + ## c_quizzes_for_this_model = c_quizzes[agreements[:, model.id]] + ## + ## one_ae_epoch(model, quiz_machine, n_epoch, c_quizzes_for_this_model, gpus[0]) time_train += int(time.perf_counter() - start_time)