tangled-llama-109m-32k-base-v0.1

logo

A pretrained language model based on the Llama model with about 109M parameters. This model has been trained on 9.7B (9,782,206,713) tokens from more than 5.2M (5,285,575) dataset rows.

This model isn't designed for immediate use but rather for Continued Pretraining and Finetuning on a downstream task. While it can handle a context length of up to 32K (32,768) tokens, it was pretrained with sequences of 2K (2048) tokens.

The objective is to streamline the cognitive or reasoning core, eliminating any redundant knowledge from the model.

loss, val_loss

val_ppl

epoch

learning_rate

lm-evaluation-harness

litgpt evaluate --tasks 'leaderboard' --out_dir 'evaluate-0/' --batch_size 4 --dtype 'bfloat16' out/pretrain/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 0.5680 ± 0.0314
- leaderboard_bbh_causal_judgement 1 none 3 acc_norm 0.5294 ± 0.0366
- leaderboard_bbh_date_understanding 1 none 3 acc_norm 0.1880 ± 0.0248
- leaderboard_bbh_disambiguation_qa 1 none 3 acc_norm 0.3240 ± 0.0297
- leaderboard_bbh_formal_fallacies 1 none 3 acc_norm 0.4720 ± 0.0316
- leaderboard_bbh_geometric_shapes 1 none 3 acc_norm 0.0280 ± 0.0105
- leaderboard_bbh_hyperbaton 1 none 3 acc_norm 0.5160 ± 0.0317
- leaderboard_bbh_logical_deduction_five_objects 1 none 3 acc_norm 0.1760 ± 0.0241
- leaderboard_bbh_logical_deduction_seven_objects 1 none 3 acc_norm 0.1360 ± 0.0217
- leaderboard_bbh_logical_deduction_three_objects 1 none 3 acc_norm 0.3480 ± 0.0302
- leaderboard_bbh_movie_recommendation 1 none 3 acc_norm 0.2280 ± 0.0266
- leaderboard_bbh_navigate 1 none 3 acc_norm 0.4800 ± 0.0317
- leaderboard_bbh_object_counting 1 none 3 acc_norm 0.0640 ± 0.0155
- leaderboard_bbh_penguins_in_a_table 1 none 3 acc_norm 0.2329 ± 0.0351
- leaderboard_bbh_reasoning_about_colored_objects 1 none 3 acc_norm 0.1240 ± 0.0209
- leaderboard_bbh_ruin_names 1 none 3 acc_norm 0.2400 ± 0.0271
- leaderboard_bbh_salient_translation_error_detection 1 none 3 acc_norm 0.1560 ± 0.0230
- leaderboard_bbh_snarks 1 none 3 acc_norm 0.4607 ± 0.0375
- leaderboard_bbh_sports_understanding 1 none 3 acc_norm 0.4560 ± 0.0316
- leaderboard_bbh_temporal_sequences 1 none 3 acc_norm 0.2320 ± 0.0268
- leaderboard_bbh_tracking_shuffled_objects_five_objects 1 none 3 acc_norm 0.2000 ± 0.0253
- 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 0.3160 ± 0.0295
- leaderboard_bbh_web_of_lies 1 none 3 acc_norm 0.5040 ± 0.0317
- leaderboard_gpqa N/A
- leaderboard_gpqa_diamond 1 none 0 acc_norm 0.1919 ± 0.0281
- leaderboard_gpqa_extended 1 none 0 acc_norm 0.2747 ± 0.0191
- leaderboard_gpqa_main 1 none 0 acc_norm 0.2589 ± 0.0207
- leaderboard_ifeval 3 none 0 inst_level_loose_acc 0.2002 ± N/A
none 0 inst_level_strict_acc 0.1871 ± N/A
none 0 prompt_level_loose_acc 0.1072 ± 0.0133
none 0 prompt_level_strict_acc 0.0998 ± 0.0129
- leaderboard_math_hard N/A
- leaderboard_math_algebra_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_counting_and_prob_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_geometry_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_intermediate_algebra_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_num_theory_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_prealgebra_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_math_precalculus_hard 1 none 4 exact_match 0.0000 ± 0
- leaderboard_mmlu_pro 0.1 none 5 acc 0.1096 ± 0.0028
- leaderboard_musr N/A
- leaderboard_musr_murder_mysteries 1 none 0 acc_norm 0.4800 ± 0.0317
- leaderboard_musr_object_placements 1 none 0 acc_norm 0.2930 ± 0.0285
- leaderboard_musr_team_allocation 1 none 0 acc_norm 0.3360 ± 0.0299
litgpt evaluate --tasks 'hellaswag,gsm8k,truthfulqa_mc2,mmlu,winogrande,arc_challenge' --out_dir 'evaluate-1/' --batch_size 4 --dtype 'bfloat16' out/pretrain/final/
Tasks Version Filter n-shot Metric Value Stderr
arc_challenge 1 none 0 acc 0.2082 ± 0.0119
none 0 acc_norm 0.2474 ± 0.0126
gsm8k 3 flexible-extract 5 exact_match 0.0106 ± 0.0028
strict-match 5 exact_match 0.0008 ± 0.0008
hellaswag 1 none 0 acc 0.2766 ± 0.0045
none 0 acc_norm 0.2926 ± 0.0045
mmlu 2 none acc 0.2349 ± 0.0036
- humanities 2 none acc 0.2461 ± 0.0063
- formal_logic 1 none 0 acc 0.2698 ± 0.0397
- high_school_european_history 1 none 0 acc 0.2000 ± 0.0312
- high_school_us_history 1 none 0 acc 0.2549 ± 0.0306
- high_school_world_history 1 none 0 acc 0.2616 ± 0.0286
- international_law 1 none 0 acc 0.2479 ± 0.0394
- jurisprudence 1 none 0 acc 0.2593 ± 0.0424
- logical_fallacies 1 none 0 acc 0.2638 ± 0.0346
- moral_disputes 1 none 0 acc 0.2457 ± 0.0232
- moral_scenarios 1 none 0 acc 0.2458 ± 0.0144
- philosophy 1 none 0 acc 0.1833 ± 0.0220
- prehistory 1 none 0 acc 0.2315 ± 0.0235
- professional_law 1 none 0 acc 0.2503 ± 0.0111
- world_religions 1 none 0 acc 0.3216 ± 0.0358
- other 2 none acc 0.2391 ± 0.0076
- business_ethics 1 none 0 acc 0.2900 ± 0.0456
- clinical_knowledge 1 none 0 acc 0.2377 ± 0.0262
- college_medicine 1 none 0 acc 0.2197 ± 0.0316
- global_facts 1 none 0 acc 0.2100 ± 0.0409
- human_aging 1 none 0 acc 0.2960 ± 0.0306
- management 1 none 0 acc 0.1748 ± 0.0376
- marketing 1 none 0 acc 0.2949 ± 0.0299
- medical_genetics 1 none 0 acc 0.2700 ± 0.0446
- miscellaneous 1 none 0 acc 0.2222 ± 0.0149
- nutrition 1 none 0 acc 0.2092 ± 0.0233
- professional_accounting 1 none 0 acc 0.2518 ± 0.0259
- professional_medicine 1 none 0 acc 0.1949 ± 0.0241
- virology 1 none 0 acc 0.3012 ± 0.0357
- social sciences 2 none acc 0.2246 ± 0.0075
- econometrics 1 none 0 acc 0.2807 ± 0.0423
- high_school_geography 1 none 0 acc 0.1818 ± 0.0275
- high_school_government_and_politics 1 none 0 acc 0.2176 ± 0.0298
- high_school_macroeconomics 1 none 0 acc 0.2179 ± 0.0209
- high_school_microeconomics 1 none 0 acc 0.2101 ± 0.0265
- high_school_psychology 1 none 0 acc 0.2000 ± 0.0171
- human_sexuality 1 none 0 acc 0.2519 ± 0.0381
- professional_psychology 1 none 0 acc 0.2516 ± 0.0176
- public_relations 1 none 0 acc 0.2182 ± 0.0396
- security_studies 1 none 0 acc 0.1959 ± 0.0254
- sociology 1 none 0 acc 0.2488 ± 0.0306
- us_foreign_policy 1 none 0 acc 0.2800 ± 0.0451
- stem 2 none acc 0.2239 ± 0.0074
- abstract_algebra 1 none 0 acc 0.1800 ± 0.0386
- anatomy 1 none 0 acc 0.1778 ± 0.0330
- astronomy 1 none 0 acc 0.1974 ± 0.0324
- college_biology 1 none 0 acc 0.2569 ± 0.0365
- college_chemistry 1 none 0 acc 0.2400 ± 0.0429
- college_computer_science 1 none 0 acc 0.2400 ± 0.0429
- college_mathematics 1 none 0 acc 0.2400 ± 0.0429
- college_physics 1 none 0 acc 0.2255 ± 0.0416
- computer_security 1 none 0 acc 0.2700 ± 0.0446
- conceptual_physics 1 none 0 acc 0.2468 ± 0.0282
- electrical_engineering 1 none 0 acc 0.2552 ± 0.0363
- elementary_mathematics 1 none 0 acc 0.2407 ± 0.0220
- high_school_biology 1 none 0 acc 0.1710 ± 0.0214
- high_school_chemistry 1 none 0 acc 0.1724 ± 0.0266
- high_school_computer_science 1 none 0 acc 0.2600 ± 0.0441
- high_school_mathematics 1 none 0 acc 0.2519 ± 0.0265
- high_school_physics 1 none 0 acc 0.1457 ± 0.0288
- high_school_statistics 1 none 0 acc 0.2083 ± 0.0277
- machine_learning 1 none 0 acc 0.3571 ± 0.0455
truthfulqa_mc2 2 none 0 acc 0.4506 ± 0.0161
winogrande 1 none 0 acc 0.5288 ± 0.0140
Groups Version Filter n-shot Metric Value Stderr
mmlu 2 none acc 0.2349 ± 0.0036
- humanities 2 none acc 0.2461 ± 0.0063
- other 2 none acc 0.2391 ± 0.0076
- social sciences 2 none acc 0.2246 ± 0.0075
- stem 2 none acc 0.2239 ± 0.0074
Downloads last month
54
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Datasets used to train tangledgroup/tangled-llama-109m-32k-base-v0.1