Marx-3B-V2 / README.md
acrastt's picture
Update README.md
cde85c3
|
raw
history blame
17.2 kB
metadata
license: apache-2.0
datasets:
  - totally-not-an-llm/EverythingLM-data-V2-sharegpt
language:
  - en
library_name: transformers

Buy Me A Coffee

This is OpenLLaMA 3B V2 finetuned on EverythingLM Data V2(ShareGPT format) for 2 epochs.

Prompt template:

### HUMAN:
{prompt}

### RESPONSE:
<leave a newline for the model to answer>

q4_1 GGML quant available here.
q4_1 GGUF quant available here.

Benchmarks:

results
hendrycksTest-moral_disputes
acc0.28612716763
acc_stderr0.0243321467791
acc_norm0.28612716763
acc_norm_stderr0.0243321467791
hendrycksTest-nutrition
acc0.287581699346
acc_stderr0.0259178061171
acc_norm0.287581699346
acc_norm_stderr0.0259178061171
hendrycksTest-prehistory
acc0.311728395062
acc_stderr0.0257731111696
acc_norm0.311728395062
acc_norm_stderr0.0257731111696
hendrycksTest-business_ethics
acc0.3
acc_stderr0.0460566186472
acc_norm0.3
acc_norm_stderr0.0460566186472
hendrycksTest-professional_medicine
acc0.180147058824
acc_stderr0.0233451636165
acc_norm0.180147058824
acc_norm_stderr0.0233451636165
hendrycksTest-international_law
acc0.388429752066
acc_stderr0.0444927035007
acc_norm0.388429752066
acc_norm_stderr0.0444927035007
hendrycksTest-high_school_world_history
acc0.286919831224
acc_stderr0.0294437730226
acc_norm0.286919831224
acc_norm_stderr0.0294437730226
hendrycksTest-management
acc0.330097087379
acc_stderr0.0465614711001
acc_norm0.330097087379
acc_norm_stderr0.0465614711001
hendrycksTest-college_biology
acc0.256944444444
acc_stderr0.0365394696944
acc_norm0.256944444444
acc_norm_stderr0.0365394696944
hendrycksTest-high_school_statistics
acc0.25462962963
acc_stderr0.02971127586
acc_norm0.25462962963
acc_norm_stderr0.02971127586
hendrycksTest-college_medicine
acc0.231213872832
acc_stderr0.0321473730203
acc_norm0.231213872832
acc_norm_stderr0.0321473730203
hendrycksTest-conceptual_physics
acc0.276595744681
acc_stderr0.0292418838696
acc_norm0.276595744681
acc_norm_stderr0.0292418838696
hendrycksTest-electrical_engineering
acc0.206896551724
acc_stderr0.0337567244956
acc_norm0.206896551724
acc_norm_stderr0.0337567244956
hendrycksTest-high_school_computer_science
acc0.33
acc_stderr0.0472581562625
acc_norm0.33
acc_norm_stderr0.0472581562625
hendrycksTest-us_foreign_policy
acc0.32
acc_stderr0.0468826172262
acc_norm0.32
acc_norm_stderr0.0468826172262
hendrycksTest-high_school_chemistry
acc0.28078817734
acc_stderr0.0316185633536
acc_norm0.28078817734
acc_norm_stderr0.0316185633536
hendrycksTest-high_school_government_and_politics
acc0.253886010363
acc_stderr0.0314102478057
acc_norm0.253886010363
acc_norm_stderr0.0314102478057
hendrycksTest-high_school_macroeconomics
acc0.274358974359
acc_stderr0.0226227657675
acc_norm0.274358974359
acc_norm_stderr0.0226227657675
hendrycksTest-college_chemistry
acc0.22
acc_stderr0.0416333199893
acc_norm0.22
acc_norm_stderr0.0416333199893
hendrycksTest-miscellaneous
acc0.315453384419
acc_stderr0.0166175017388
acc_norm0.315453384419
acc_norm_stderr0.0166175017388
hendrycksTest-high_school_psychology
acc0.28623853211
acc_stderr0.0193794366289
acc_norm0.28623853211
acc_norm_stderr0.0193794366289
hendrycksTest-machine_learning
acc0.214285714286
acc_stderr0.0389464112004
acc_norm0.214285714286
acc_norm_stderr0.0389464112004
truthfulqa_mc
mc1_stderr0.0150006743736
mc2_stderr0.0138986280365
mc20.38597579296
mc10.2423500612
hendrycksTest-marketing
acc0.290598290598
acc_stderr0.0297450485727
acc_norm0.290598290598
acc_norm_stderr0.0297450485727
hendrycksTest-world_religions
acc0.327485380117
acc_stderr0.0359933577146
acc_norm0.327485380117
acc_norm_stderr0.0359933577146
hendrycksTest-logical_fallacies
acc0.233128834356
acc_stderr0.0332201579578
acc_norm0.233128834356
acc_norm_stderr0.0332201579578
hendrycksTest-security_studies
acc0.326530612245
acc_stderr0.0300210562384
acc_norm0.326530612245
acc_norm_stderr0.0300210562384
hendrycksTest-college_mathematics
acc0.31
acc_stderr0.0464823198712
acc_norm0.31
acc_norm_stderr0.0464823198712
hendrycksTest-human_aging
acc0.381165919283
acc_stderr0.0325962511842
acc_norm0.381165919283
acc_norm_stderr0.0325962511842
hendrycksTest-formal_logic
acc0.261904761905
acc_stderr0.0393253768039
acc_norm0.261904761905
acc_norm_stderr0.0393253768039
hendrycksTest-sociology
acc0.258706467662
acc_stderr0.0309659031236
acc_norm0.258706467662
acc_norm_stderr0.0309659031236
hendrycksTest-public_relations
acc0.309090909091
acc_stderr0.044262946482
acc_norm0.309090909091
acc_norm_stderr0.044262946482
hendrycksTest-high_school_physics
acc0.304635761589
acc_stderr0.0375794992294
acc_norm0.304635761589
acc_norm_stderr0.0375794992294
hendrycksTest-human_sexuality
acc0.198473282443
acc_stderr0.0349814938546
acc_norm0.198473282443
acc_norm_stderr0.0349814938546
hendrycksTest-clinical_knowledge
acc0.264150943396
acc_stderr0.0271342916287
acc_norm0.264150943396
acc_norm_stderr0.0271342916287
hendrycksTest-high_school_biology
acc0.241935483871
acc_stderr0.024362599693
acc_norm0.241935483871
acc_norm_stderr0.024362599693
hendrycksTest-high_school_us_history
acc0.230392156863
acc_stderr0.0295542926057
acc_norm0.230392156863
acc_norm_stderr0.0295542926057
hendrycksTest-high_school_mathematics
acc0.262962962963
acc_stderr0.0268420578738
acc_norm0.262962962963
acc_norm_stderr0.0268420578738
hendrycksTest-medical_genetics
acc0.29
acc_stderr0.0456048021572
acc_norm0.29
acc_norm_stderr0.0456048021572
hendrycksTest-moral_scenarios
acc0.212290502793
acc_stderr0.0136766446858
acc_norm0.212290502793
acc_norm_stderr0.0136766446858
hendrycksTest-high_school_microeconomics
acc0.264705882353
acc_stderr0.0286574912851
acc_norm0.264705882353
acc_norm_stderr0.0286574912851
hendrycksTest-philosophy
acc0.27652733119
acc_stderr0.0254038329782
acc_norm0.27652733119
acc_norm_stderr0.0254038329782
hendrycksTest-virology
acc0.301204819277
acc_stderr0.0357160923005
acc_norm0.301204819277
acc_norm_stderr0.0357160923005
hendrycksTest-econometrics
acc0.263157894737
acc_stderr0.0414243971949
acc_norm0.263157894737
acc_norm_stderr0.0414243971949
hendrycksTest-abstract_algebra
acc0.23
acc_stderr0.0422952584682
acc_norm0.23
acc_norm_stderr0.0422952584682
hendrycksTest-college_physics
acc0.274509803922
acc_stderr0.0444052190618
acc_norm0.274509803922
acc_norm_stderr0.0444052190618
hendrycksTest-astronomy
acc0.328947368421
acc_stderr0.0382342896993
acc_norm0.328947368421
acc_norm_stderr0.0382342896993
hendrycksTest-global_facts
acc0.35
acc_stderr0.0479372485441
acc_norm0.35
acc_norm_stderr0.0479372485441
hendrycksTest-computer_security
acc0.36
acc_stderr0.0482418151324
acc_norm0.36
acc_norm_stderr0.0482418151324
hendrycksTest-high_school_geography
acc0.287878787879
acc_stderr0.032258835123
acc_norm0.287878787879
acc_norm_stderr0.032258835123
hendrycksTest-high_school_european_history
acc0.29696969697
acc_stderr0.0356796977227
acc_norm0.29696969697
acc_norm_stderr0.0356796977227
hendrycksTest-professional_psychology
acc0.279411764706
acc_stderr0.0181528710515
acc_norm0.279411764706
acc_norm_stderr0.0181528710515
hellaswag
acc0.551384186417
acc_stderr0.00496336208528
acc_norm0.725751842262
acc_norm_stderr0.00445222854104
hendrycksTest-professional_accounting
acc0.265957446809
acc_stderr0.0263580656989
acc_norm0.265957446809
acc_norm_stderr0.0263580656989
hendrycksTest-college_computer_science
acc0.26
acc_stderr0.0440844002277
acc_norm0.26
acc_norm_stderr0.0440844002277
hendrycksTest-professional_law
acc0.256844850065
acc_stderr0.0111584558531
acc_norm0.256844850065
acc_norm_stderr0.0111584558531
hendrycksTest-jurisprudence
acc0.231481481481
acc_stderr0.0407749470925
acc_norm0.231481481481
acc_norm_stderr0.0407749470925
hendrycksTest-elementary_mathematics
acc0.269841269841
acc_stderr0.0228608383092
acc_norm0.269841269841
acc_norm_stderr0.0228608383092
hendrycksTest-anatomy
acc0.296296296296
acc_stderr0.039446241625
acc_norm0.296296296296
acc_norm_stderr0.039446241625
arc_challenge
acc0.389931740614
acc_stderr0.0142529598489
acc_norm0.430887372014
acc_norm_stderr0.0144711333926