--- license: apache-2.0 datasets: - LDJnr/Puffin language: - en library_name: transformers pipeline_tag: text-generation --- This is [OpenLLaMA 3B V2](https://huggingface.co/openlm-research/open_llama_3b_v2) finetuned on [Puffin](https://huggingface.co/datasets/LDJnr/Puffin) for 1 epochs. Prompt template: ``` ### HUMAN: {prompt} ### RESPONSE: ``` GGML quants available [here](https://huggingface.co/TheBloke/Griffin-3B-GGML).
GPTQ quants available [here](https://huggingface.co/TheBloke/Griffin-3B-GPTQ). Note: Don't expect this model to be good, I was just starting out to finetune. So don't roast me please! Benchmarks:
hellaswag
acc0.5415255925114519
acc_stderr0.004972543127767884
acc_norm0.7211710814578769
acc_norm_stderr0.004475067344626758
hendrycksTest-abstract_algebra
acc0.22
acc_stderr0.04163331998932269
acc_norm0.22
acc_norm_stderr0.04163331998932269
hendrycksTest-anatomy
acc0.21481481481481482
acc_stderr0.03547854198560824
acc_norm0.21481481481481482
acc_norm_stderr0.03547854198560824
hendrycksTest-astronomy
acc0.2894736842105263
acc_stderr0.03690677986137283
acc_norm0.2894736842105263
acc_norm_stderr0.03690677986137283
hendrycksTest-business_ethics
acc0.37
acc_stderr0.04852365870939099
acc_norm0.37
acc_norm_stderr0.04852365870939099
hendrycksTest-clinical_knowledge
acc0.2641509433962264
acc_stderr0.02713429162874172
acc_norm0.2641509433962264
acc_norm_stderr0.02713429162874172
hendrycksTest-college_biology
acc0.2361111111111111
acc_stderr0.03551446610810826
acc_norm0.2361111111111111
acc_norm_stderr0.03551446610810826
hendrycksTest-college_chemistry
acc0.24
acc_stderr0.04292346959909284
acc_norm0.24
acc_norm_stderr0.04292346959909284
hendrycksTest-college_computer_science
acc0.22
acc_stderr0.04163331998932269
acc_norm0.22
acc_norm_stderr0.04163331998932269
hendrycksTest-college_mathematics
acc0.3
acc_stderr0.046056618647183814
acc_norm0.3
acc_norm_stderr0.046056618647183814
hendrycksTest-college_medicine
acc0.21965317919075145
acc_stderr0.031568093627031744
acc_norm0.21965317919075145
acc_norm_stderr0.031568093627031744
hendrycksTest-college_physics
acc0.19607843137254902
acc_stderr0.03950581861179961
acc_norm0.19607843137254902
acc_norm_stderr0.03950581861179961
hendrycksTest-computer_security
acc0.33
acc_stderr0.04725815626252605
acc_norm0.33
acc_norm_stderr0.04725815626252605
hendrycksTest-conceptual_physics
acc0.3148936170212766
acc_stderr0.03036358219723817
acc_norm0.3148936170212766
acc_norm_stderr0.03036358219723817
hendrycksTest-econometrics
acc0.22807017543859648
acc_stderr0.03947152782669416
acc_norm0.22807017543859648
acc_norm_stderr0.03947152782669416
hendrycksTest-electrical_engineering
acc0.2
acc_stderr0.0333333333333333
acc_norm0.2
acc_norm_stderr0.0333333333333333
hendrycksTest-elementary_mathematics
acc0.29365079365079366
acc_stderr0.023456037383982022
acc_norm0.29365079365079366
acc_norm_stderr0.023456037383982022
hendrycksTest-formal_logic
acc0.1984126984126984
acc_stderr0.03567016675276862
acc_norm0.1984126984126984
acc_norm_stderr0.03567016675276862
hendrycksTest-global_facts
acc0.32
acc_stderr0.046882617226215034
acc_norm0.32
acc_norm_stderr0.046882617226215034
hendrycksTest-high_school_biology
acc0.23225806451612904
acc_stderr0.02402225613030824
acc_norm0.23225806451612904
acc_norm_stderr0.02402225613030824
hendrycksTest-high_school_chemistry
acc0.23645320197044334
acc_stderr0.029896114291733552
acc_norm0.23645320197044334
acc_norm_stderr0.029896114291733552
hendrycksTest-high_school_computer_science
acc0.26
acc_stderr0.044084400227680794
acc_norm0.26
acc_norm_stderr0.044084400227680794
hendrycksTest-high_school_european_history
acc0.22424242424242424
acc_stderr0.032568666616811015
acc_norm0.22424242424242424
acc_norm_stderr0.032568666616811015
hendrycksTest-high_school_geography
acc0.20707070707070707
acc_stderr0.02886977846026705
acc_norm0.20707070707070707
acc_norm_stderr0.02886977846026705
hendrycksTest-high_school_government_and_politics
acc0.22797927461139897
acc_stderr0.030276909945178274
acc_norm0.22797927461139897
acc_norm_stderr0.030276909945178274
hendrycksTest-high_school_macroeconomics
acc0.2358974358974359
acc_stderr0.021525965407408726
acc_norm0.2358974358974359
acc_norm_stderr0.021525965407408726
hendrycksTest-high_school_mathematics
acc0.25925925925925924
acc_stderr0.026719240783712163
acc_norm0.25925925925925924
acc_norm_stderr0.026719240783712163
hendrycksTest-high_school_microeconomics
acc0.2815126050420168
acc_stderr0.029213549414372167
acc_norm0.2815126050420168
acc_norm_stderr0.029213549414372167
hendrycksTest-high_school_physics
acc0.2980132450331126
acc_stderr0.037345356767871984
acc_norm0.2980132450331126
acc_norm_stderr0.037345356767871984
hendrycksTest-high_school_psychology
acc0.23669724770642203
acc_stderr0.01822407811729907
acc_norm0.23669724770642203
acc_norm_stderr0.01822407811729907
hendrycksTest-high_school_statistics
acc0.1574074074074074
acc_stderr0.02483717351824239
acc_norm0.1574074074074074
acc_norm_stderr0.02483717351824239
hendrycksTest-high_school_us_history
acc0.2647058823529412
acc_stderr0.03096451792692341
acc_norm0.2647058823529412
acc_norm_stderr0.03096451792692341
hendrycksTest-high_school_world_history
acc0.28270042194092826
acc_stderr0.02931281415395592
acc_norm0.28270042194092826
acc_norm_stderr0.02931281415395592
hendrycksTest-human_aging
acc0.39461883408071746
acc_stderr0.03280400504755291
acc_norm0.39461883408071746
acc_norm_stderr0.03280400504755291
hendrycksTest-human_sexuality
acc0.22900763358778625
acc_stderr0.036853466317118506
acc_norm0.22900763358778625
acc_norm_stderr0.036853466317118506
hendrycksTest-international_law
acc0.30578512396694213
acc_stderr0.04205953933884124
acc_norm0.30578512396694213
acc_norm_stderr0.04205953933884124
hendrycksTest-jurisprudence
acc0.32407407407407407
acc_stderr0.04524596007030048
acc_norm0.32407407407407407
acc_norm_stderr0.04524596007030048
hendrycksTest-logical_fallacies
acc0.22699386503067484
acc_stderr0.03291099578615769
acc_norm0.22699386503067484
acc_norm_stderr0.03291099578615769
hendrycksTest-machine_learning
acc0.2857142857142857
acc_stderr0.04287858751340456
acc_norm0.2857142857142857
acc_norm_stderr0.04287858751340456
hendrycksTest-management
acc0.27184466019417475
acc_stderr0.044052680241409216
acc_norm0.27184466019417475
acc_norm_stderr0.044052680241409216
hendrycksTest-marketing
acc0.27350427350427353
acc_stderr0.02920254015343117
acc_norm0.27350427350427353
acc_norm_stderr0.02920254015343117
hendrycksTest-medical_genetics
acc0.23
acc_stderr0.04229525846816505
acc_norm0.23
acc_norm_stderr0.04229525846816505
hendrycksTest-miscellaneous
acc0.2937420178799489
acc_stderr0.016287759388491672
acc_norm0.2937420178799489
acc_norm_stderr0.016287759388491672
hendrycksTest-moral_disputes
acc0.2774566473988439
acc_stderr0.024105712607754307
acc_norm0.2774566473988439
acc_norm_stderr0.024105712607754307
hendrycksTest-moral_scenarios
acc0.24916201117318434
acc_stderr0.014465893829859933
acc_norm0.24916201117318434
acc_norm_stderr0.014465893829859933
hendrycksTest-nutrition
acc0.24836601307189543
acc_stderr0.024739981355113596
acc_norm0.24836601307189543
acc_norm_stderr0.024739981355113596
hendrycksTest-philosophy
acc0.2733118971061093
acc_stderr0.02531176597542612
acc_norm0.2733118971061093
acc_norm_stderr0.02531176597542612
hendrycksTest-prehistory
acc0.2623456790123457
acc_stderr0.024477222856135107
acc_norm0.2623456790123457
acc_norm_stderr0.024477222856135107
hendrycksTest-professional_accounting
acc0.2801418439716312
acc_stderr0.026789172351140228
acc_norm0.2801418439716312
acc_norm_stderr0.026789172351140228
hendrycksTest-professional_law
acc0.24185136897001303
acc_stderr0.010936550813827065
acc_norm0.24185136897001303
acc_norm_stderr0.010936550813827065
hendrycksTest-professional_medicine
acc0.20220588235294118
acc_stderr0.02439819298665492
acc_norm0.20220588235294118
acc_norm_stderr0.02439819298665492
hendrycksTest-professional_psychology
acc0.25980392156862747
acc_stderr0.017740899509177795
acc_norm0.25980392156862747
acc_norm_stderr0.017740899509177795
hendrycksTest-public_relations
acc0.34545454545454546
acc_stderr0.04554619617541054
acc_norm0.34545454545454546
acc_norm_stderr0.04554619617541054
hendrycksTest-security_studies
acc0.23673469387755103
acc_stderr0.02721283588407316
acc_norm0.23673469387755103
acc_norm_stderr0.02721283588407316
hendrycksTest-sociology
acc0.23880597014925373
acc_stderr0.030147775935409224
acc_norm0.23880597014925373
acc_norm_stderr0.030147775935409224
hendrycksTest-us_foreign_policy
acc0.29
acc_stderr0.04560480215720684
acc_norm0.29
acc_norm_stderr0.04560480215720684
hendrycksTest-virology
acc0.3132530120481928
acc_stderr0.036108050180310235
acc_norm0.3132530120481928
acc_norm_stderr0.036108050180310235
hendrycksTest-world_religions
acc0.34502923976608185
acc_stderr0.036459813773888065
acc_norm0.34502923976608185
acc_norm_stderr0.036459813773888065
truthfulqa_mc
mc10.23990208078335373
mc1_stderr0.014948812679062133
mc20.38208865449224466
mc2_stderr0.013866257677161405
arc_challenge
acc0.3856655290102389
acc_stderr0.014224250973257177
acc_norm0.41723549488054607
acc_norm_stderr0.014409825518403079