--- license: mit pipeline_tag: text-generation library_name: transformers language: [ 'en', 'am', 'ar', 'as', 'az', 'be', 'bg', 'bn', 'br', 'bs', 'ca', 'cs', 'cy', 'da', 'de', 'el', 'eo', 'es', 'et', 'eu', 'fa', 'ff', 'fi', 'fr', 'fy', 'ga', 'gd', 'gl', 'gn', 'gu', 'ha', 'he', 'hi', 'hr', 'ht', 'hu', 'hy', 'id', 'ig', 'is', 'it', 'ja', 'jv', 'ka', 'kk', 'km', 'kn', 'ko', 'ku', 'ky', 'la', 'lg', 'li', 'ln', 'lo', 'lt', 'lv', 'mg', 'mk', 'ml', 'mn', 'mr', 'ms', 'my', 'ne', 'nl', 'no', 'ns', 'om', 'or', 'pa', 'pl', 'ps', 'pt', 'qu', 'rm', 'ro', 'ru', 'sa', 'si', 'sc', 'sd', 'sk', 'sl', 'so', 'sq', 'sr', 'ss', 'su', 'sv', 'sw', 'ta', 'te', 'th', 'tl', 'tn', 'tr', 'ug', 'uk', 'ur', 'uz', 'vi', 'wo', 'xh', 'yi', 'yo', 'zu', ] datasets: # core - base - ontocord/fineweb-permissive-multilingual-2m - distily/c4_multilingual_1M - data-silence/sumnews - xu-song/cc100-samples - badrex/llm-emoji-dataset - fblgit/simple-math - Gusarich/math-expressions-1m - neuralwork/arxiver - christopher/rosetta-code - nampdn-ai/tiny-codes - JeanKaddour/minipile # core - instruct - NousResearch/hermes-function-calling-v1 - simplescaling/s1K-1.1 # base - instruct - mlabonne/open-perfectblend - allenai/tulu-3-sft-mixture - rombodawg/Everything_Instruct_Multilingual # base - reason - open-r1/OpenR1-Math-220k - open-thoughts/OpenThoughts-114k - cognitivecomputations/dolphin-r1 - simplescaling/s1K-1.1 tags: - chat - core - base - instruct - reason --- # tangled-alpha-0.14-core ![logo](./misc/logo.jpg) ```bash time python -B prepare_base_datasets.py ``` ``` i=0, min_len=0, max_len=1073741824, block_size=8193, chunk_size=16386000, len(dataset)=1496631, len(dataset) * block_size=12261897783 Total number of tokens in the optimized dataset '../base-data-0-0-1073741824-8193-2000' is 12261897783 i=1, min_len=8193, max_len=16385, block_size=16385, chunk_size=16385000, len(dataset)=78802, len(dataset) * block_size=1291170770 Total number of tokens in the optimized dataset '../base-data-1-8193-16385-16385-1000' is 1291170770 i=2, min_len=16385, max_len=32769, block_size=32769, chunk_size=16384500, len(dataset)=23511, len(dataset) * block_size=770431959 Total number of tokens in the optimized dataset '../base-data-2-16385-32769-32769-500' is 770431959 i=3, min_len=32769, max_len=65537, block_size=65537, chunk_size=16384250, len(dataset)=5128, len(dataset) * block_size=336073736 Total number of tokens in the optimized dataset '../base-data-3-32769-65537-65537-250' is 336073736 i=4, min_len=65537, max_len=131073, block_size=131073, chunk_size=16384125, len(dataset)=1169, len(dataset) * block_size=153224337 Total number of tokens in the optimized dataset '../base-data-4-65537-131073-131073-125' is 153224337 46G ../base-data-0-0-1073741824-8193-2000 4.9G ../base-data-1-8193-16385-16385-1000 2.9G ../base-data-2-16385-32769-32769-500 1.3G ../base-data-3-32769-65537-65537-250 589M ../base-data-4-65537-131073-131073-125 ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt pretrain --config pretrain_base_model_0.yaml ``` ``` ``` Backup `wandb`: ```bash mv wandb wandb-pretrain-base-0 ``` Copy config: ```bash cp ../config-0.json ../out/pretrain-base-0/final/config.json ``` Chat with model: ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt chat ../out/pretrain-base-0/final ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True time litgpt evaluate --tasks 'leaderboard' --out_dir '../evaluate/pretrain-base-0/leaderboard/' --batch_size '4' --dtype 'bfloat16' '../out/pretrain-base-0/final' ``` ``` | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |-----------------------------------------------------------|-------|------|-----:|-----------------------|---|-----:|---|------| |leaderboard | N/A| | | | | | | | | - leaderboard_bbh | N/A| | | | | | | | | - leaderboard_bbh_boolean_expressions | 1|none | 3|acc_norm |↑ |0.4560|± |0.0316| | - leaderboard_bbh_causal_judgement | 1|none | 3|acc_norm |↑ |0.5187|± |0.0366| | - leaderboard_bbh_date_understanding | 1|none | 3|acc_norm |↑ |0.2000|± |0.0253| | - leaderboard_bbh_disambiguation_qa | 1|none | 3|acc_norm |↑ |0.3400|± |0.0300| | - leaderboard_bbh_formal_fallacies | 1|none | 3|acc_norm |↑ |0.4680|± |0.0316| | - leaderboard_bbh_geometric_shapes | 1|none | 3|acc_norm |↑ |0.0880|± |0.0180| | - leaderboard_bbh_hyperbaton | 1|none | 3|acc_norm |↑ |0.5160|± |0.0317| | - leaderboard_bbh_logical_deduction_five_objects | 1|none | 3|acc_norm |↑ |0.1880|± |0.0248| | - leaderboard_bbh_logical_deduction_seven_objects | 1|none | 3|acc_norm |↑ |0.1440|± |0.0222| | - leaderboard_bbh_logical_deduction_three_objects | 1|none | 3|acc_norm |↑ |0.3360|± |0.0299| | - leaderboard_bbh_movie_recommendation | 1|none | 3|acc_norm |↑ |0.2680|± |0.0281| | - leaderboard_bbh_navigate | 1|none | 3|acc_norm |↑ |0.5800|± |0.0313| | - leaderboard_bbh_object_counting | 1|none | 3|acc_norm |↑ |0.0560|± |0.0146| | - leaderboard_bbh_penguins_in_a_table | 1|none | 3|acc_norm |↑ |0.2055|± |0.0336| | - leaderboard_bbh_reasoning_about_colored_objects | 1|none | 3|acc_norm |↑ |0.1400|± |0.0220| | - leaderboard_bbh_ruin_names | 1|none | 3|acc_norm |↑ |0.2160|± |0.0261| | - leaderboard_bbh_salient_translation_error_detection | 1|none | 3|acc_norm |↑ |0.1120|± |0.0200| | - leaderboard_bbh_snarks | 1|none | 3|acc_norm |↑ |0.5056|± |0.0376| | - leaderboard_bbh_sports_understanding | 1|none | 3|acc_norm |↑ |0.4800|± |0.0317| | - leaderboard_bbh_temporal_sequences | 1|none | 3|acc_norm |↑ |0.2840|± |0.0286| | - leaderboard_bbh_tracking_shuffled_objects_five_objects | 1|none | 3|acc_norm |↑ |0.2400|± |0.0271| | - leaderboard_bbh_tracking_shuffled_objects_seven_objects| 1|none | 3|acc_norm |↑ |0.1520|± |0.0228| | - leaderboard_bbh_tracking_shuffled_objects_three_objects| 1|none | 3|acc_norm |↑ |0.3320|± |0.0298| | - leaderboard_bbh_web_of_lies | 1|none | 3|acc_norm |↑ |0.4880|± |0.0317| | - leaderboard_gpqa | N/A| | | | | | | | | - leaderboard_gpqa_diamond | 1|none | 0|acc_norm |↑ |0.2071|± |0.0289| | - leaderboard_gpqa_extended | 1|none | 0|acc_norm |↑ |0.2637|± |0.0189| | - leaderboard_gpqa_main | 1|none | 0|acc_norm |↑ |0.2612|± |0.0208| | - leaderboard_ifeval | 3|none | 0|inst_level_loose_acc |↑ |0.2590|± | N/A| | | |none | 0|inst_level_strict_acc |↑ |0.2494|± | N/A| | | |none | 0|prompt_level_loose_acc |↑ |0.1497|± |0.0154| | | |none | 0|prompt_level_strict_acc|↑ |0.1405|± |0.0150| | - leaderboard_math_hard | N/A| | | | | | | | | - leaderboard_math_algebra_hard | 2|none | 4|exact_match |↑ |0.0008|± |0.0008| | - leaderboard_math_counting_and_prob_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_math_geometry_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_math_intermediate_algebra_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_math_num_theory_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_math_prealgebra_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_math_precalculus_hard | 2|none | 4|exact_match |↑ |0.0000|± | 0| | - leaderboard_mmlu_pro | 0.1|none | 5|acc |↑ |0.1112|± |0.0029| | - leaderboard_musr | N/A| | | | | | | | | - leaderboard_musr_murder_mysteries | 1|none | 0|acc_norm |↑ |0.5240|± |0.0316| | - leaderboard_musr_object_placements | 1|none | 0|acc_norm |↑ |0.2578|± |0.0274| | - leaderboard_musr_team_allocation | 1|none | 0|acc_norm |↑ |0.3960|± |0.0310| ``` ```bash litgpt convert_pretrained_checkpoint ../out/pretrain-base-0/final ../out/pretrain-base-0/checkpoint ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt pretrain --config pretrain_base_model_1.yaml ``` ```bash litgpt convert_pretrained_checkpoint ../out/pretrain-base-1/final ../out/pretrain-base-1/checkpoint ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt pretrain --config pretrain_base_model_2.yaml ``` ```bash litgpt convert_pretrained_checkpoint ../out/pretrain-base-2/final ../out/pretrain-base-2/checkpoint ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True litgpt pretrain --config pretrain_base_model_3.yaml ``` ```bash CUDA_VISIBLE_DEVICES=0 CUDA_LAUNCH_BLOCKING=0 PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True time litgpt evaluate --tasks 'leaderboard' --out_dir '../evaluate/pretrain-base-3/leaderboard/' --batch_size '4' --dtype 'bfloat16' '../out/pretrain-base-3/final' ``` ``` ```