From 1a03b01a527405037696d68aadff005be2cf5f8a Mon Sep 17 00:00:00 2001 From: 51616 Date: Thu, 3 Jul 2025 11:02:46 +0000 Subject: [PATCH] small adjustment data gen (remove empty) --- configs/qa_short_ctx_self_gen_no_fw_qa.yaml | 8 ++++---- data/generate_fw_edu_qa_v2.py | 8 ++++++-- data/generate_fw_edu_qa_v2_repeat.py | 8 ++++++-- data/self_generate_qa.py | 2 +- 4 files changed, 17 insertions(+), 9 deletions(-) diff --git a/configs/qa_short_ctx_self_gen_no_fw_qa.yaml b/configs/qa_short_ctx_self_gen_no_fw_qa.yaml index de4165b..030e683 100644 --- a/configs/qa_short_ctx_self_gen_no_fw_qa.yaml +++ b/configs/qa_short_ctx_self_gen_no_fw_qa.yaml @@ -39,10 +39,10 @@ target_modules: # data train_ds_names: - - self_gen/google/gemma-2-2b-it/squad_compact - - self_gen/google/gemma-2-2b-it/pwc_compact - - self_gen/google/gemma-2-2b-it/ropes_compact - - self_gen/google/gemma-2-2b-it/drop_compact + - self_gen/google/gemma-2-2b-it_temp_0.0/squad_compact + - self_gen/google/gemma-2-2b-it_temp_0.0/pwc_compact + - self_gen/google/gemma-2-2b-it_temp_0.0/ropes_compact + - self_gen/google/gemma-2-2b-it_temp_0.0/drop_compact val_ds_names: - squad diff --git a/data/generate_fw_edu_qa_v2.py b/data/generate_fw_edu_qa_v2.py index d3a0505..0a5b9bd 100644 --- a/data/generate_fw_edu_qa_v2.py +++ b/data/generate_fw_edu_qa_v2.py @@ -101,14 +101,18 @@ def postprocess_qa_pairs(res_txt: str): has_left_over = n_gen_pairs < len(questions) or n_gen_pairs < len(answers) for i in range(n_gen_pairs): response = answers[i].strip() + question = questions[i].strip() + if not response or not question: + print(f"Skipping empty question or answer at index {i}") + continue if (not has_left_over) and (i == n_gen_pairs - 1): response, skip = check_should_skip(response, vllm_model) if skip: print(f"Skipping due to missing stop string") n_skips += 1 continue - out_q.append(questions[i].strip()) - out_a.append(answers[i].strip()) + out_q.append(question.strip()) + out_a.append(response.strip()) print(f"Skipped {n_skips} responses due to missing stop strings") return out_q, out_a diff --git a/data/generate_fw_edu_qa_v2_repeat.py b/data/generate_fw_edu_qa_v2_repeat.py index 6c70103..7a85fe5 100644 --- a/data/generate_fw_edu_qa_v2_repeat.py +++ b/data/generate_fw_edu_qa_v2_repeat.py @@ -108,14 +108,18 @@ def postprocess_qa_pairs(res_txt: str): has_left_over = n_gen_pairs < len(questions) or n_gen_pairs < len(answers) for i in range(n_gen_pairs): response = answers[i].strip() + question = questions[i].strip() + if not response or not question: + print(f"Skipping empty question or answer at index {i}") + continue if (not has_left_over) and (i == n_gen_pairs - 1): response, skip = check_should_skip(response, vllm_model) if skip: print(f"Skipping due to missing stop string") n_skips += 1 continue - out_q.append(questions[i].strip()) - out_a.append(answers[i].strip()) + out_q.append(question.strip()) + out_a.append(response.strip()) print(f"Skipped {n_skips} responses due to missing stop strings") return out_q, out_a diff --git a/data/self_generate_qa.py b/data/self_generate_qa.py index 0a7ed33..e7ecd2c 100644 --- a/data/self_generate_qa.py +++ b/data/self_generate_qa.py @@ -237,7 +237,7 @@ def self_generate( print("=" * 80) print(f"Generated {len(samples)} samples") - random.shuffle(samples) + # random.shuffle(samples) # Save results if not args.debug: