metadata
language:
- en
license: apache-2.0
tags:
- maths
- gpt2
- mathgpt2
- trl
- sft
datasets:
- meta-math/MetaMathQA
- ArtifactAI/arxiv-math-instruct-50k
pipeline_tag: text-generation
widget:
- text: Which motion is formed by an incident particle?
example_title: Example 1
- text: What type of diffusional modeling is used for diffusion?
example_title: Example 2
model-index:
- name: math_gpt2_sft
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: AI2 Reasoning Challenge (25-Shot)
type: ai2_arc
config: ARC-Challenge
split: test
args:
num_few_shot: 25
metrics:
- type: acc_norm
value: 22.87
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: HellaSwag (10-Shot)
type: hellaswag
split: validation
args:
num_few_shot: 10
metrics:
- type: acc_norm
value: 30.41
name: normalized accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU (5-Shot)
type: cais/mmlu
config: all
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 25.06
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: TruthfulQA (0-shot)
type: truthful_qa
config: multiple_choice
split: validation
args:
num_few_shot: 0
metrics:
- type: mc2
value: 37.62
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: Winogrande (5-shot)
type: winogrande
config: winogrande_xl
split: validation
args:
num_few_shot: 5
metrics:
- type: acc
value: 51.54
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
- task:
type: text-generation
name: Text Generation
dataset:
name: GSM8k (5-shot)
type: gsm8k
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 0.68
name: accuracy
source:
url: >-
https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Sharathhebbar24/math_gpt2_sft
name: Open LLM Leaderboard
This model is a finetuned version of Sharathhebbar24/math_gpt2 using meta-math/MetaMathQA
Model description
GPT-2 is a transformers model pre-trained on a very large corpus of English data in a self-supervised fashion. This means it was pre-trained on the raw texts only, with no humans labeling them in any way (which is why it can use lots of publicly available data) with an automatic process to generate inputs and labels from those texts. More precisely, it was trained to guess the next word in sentences.
More precisely, inputs are sequences of continuous text of a certain length and the targets are the same sequence,
shifting one token (word or piece of word) to the right. The model uses a masking mechanism to make sure the
predictions for the token i only use the inputs from 1 to i but not the future tokens.
This way, the model learns an inner representation of the English language that can then be used to extract features useful for downstream tasks. The model is best at what it was trained for, however, which is generating texts from a prompt.
To use this model
>>> from transformers import AutoTokenizer, AutoModelForCausalLM
>>> model_name = "Sharathhebbar24/math_gpt2_sft"
>>> model = AutoModelForCausalLM.from_pretrained(model_name)
>>> tokenizer = AutoTokenizer.from_pretrained(model_name)
>>> def generate_text(prompt):
>>> inputs = tokenizer.encode(prompt, return_tensors='pt')
>>> outputs = model.generate(inputs, max_length=64, pad_token_id=tokenizer.eos_token_id)
>>> generated = tokenizer.decode(outputs[0], skip_special_tokens=True)
>>> return generated[:generated.rfind(".")+1]
>>> prompt = "Gracie and Joe are choosing numbers on the complex plane. Joe chooses the point $1+2i$. Gracie chooses $-1+i$. How far apart are Gracie and Joe's points?"
>>> res = generate_text(prompt)
>>> res
Benchmark / Evaluation
| Model | Average | ARC | HellaSwag | MMLU | TruthfulQA | Winogrande | GSM8k |
|---|---|---|---|---|---|---|---|
| Sharathhebbar24/math_gpt2_sft | 28.503 | 22.87 | 30.41 | 25.06 | 37.62 | 51.54 | 0.68 |
{
"all": {
"acc": 0.25082189621988066,
"acc_stderr": 0.030526589726831692,
"acc_norm": 0.25112870356236633,
"acc_norm_stderr": 0.03129390389566968,
"mc1": 0.24112607099143207,
"mc1_stderr": 0.014974827279752334,
"mc2": 0.3762297840067963,
"mc2_stderr": 0.01445991036363257
},
"harness|arc:challenge|25": {
"acc": 0.20563139931740615,
"acc_stderr": 0.01181074526074258,
"acc_norm": 0.22866894197952217,
"acc_norm_stderr": 0.012272853582540799
},
"harness|hellaswag|10": {
"acc": 0.2884883489344752,
"acc_stderr": 0.004521334761709224,
"acc_norm": 0.30412268472415854,
"acc_norm_stderr": 0.00459094683972719
},
"harness|hendrycksTest-abstract_algebra|5": {
"acc": 0.19,
"acc_stderr": 0.03942772444036625,
"acc_norm": 0.19,
"acc_norm_stderr": 0.03942772444036625
},
"harness|hendrycksTest-anatomy|5": {
"acc": 0.2074074074074074,
"acc_stderr": 0.03502553170678319,
"acc_norm": 0.2074074074074074,
"acc_norm_stderr": 0.03502553170678319
},
"harness|hendrycksTest-astronomy|5": {
"acc": 0.17763157894736842,
"acc_stderr": 0.031103182383123398,
"acc_norm": 0.17763157894736842,
"acc_norm_stderr": 0.031103182383123398
},
"harness|hendrycksTest-business_ethics|5": {
"acc": 0.19,
"acc_stderr": 0.03942772444036622,
"acc_norm": 0.19,
"acc_norm_stderr": 0.03942772444036622
},
"harness|hendrycksTest-clinical_knowledge|5": {
"acc": 0.2188679245283019,
"acc_stderr": 0.025447863825108618,
"acc_norm": 0.2188679245283019,
"acc_norm_stderr": 0.025447863825108618
},
"harness|hendrycksTest-college_biology|5": {
"acc": 0.25,
"acc_stderr": 0.03621034121889507,
"acc_norm": 0.25,
"acc_norm_stderr": 0.03621034121889507
},
"harness|hendrycksTest-college_chemistry|5": {
"acc": 0.2,
"acc_stderr": 0.04020151261036845,
"acc_norm": 0.2,
"acc_norm_stderr": 0.04020151261036845
},
"harness|hendrycksTest-college_computer_science|5": {
"acc": 0.32,
"acc_stderr": 0.046882617226215034,
"acc_norm": 0.32,
"acc_norm_stderr": 0.046882617226215034
},
"harness|hendrycksTest-college_mathematics|5": {
"acc": 0.24,
"acc_stderr": 0.042923469599092816,
"acc_norm": 0.24,
"acc_norm_stderr": 0.042923469599092816
},
"harness|hendrycksTest-college_medicine|5": {
"acc": 0.21965317919075145,
"acc_stderr": 0.031568093627031744,
"acc_norm": 0.21965317919075145,
"acc_norm_stderr": 0.031568093627031744
},
"harness|hendrycksTest-college_physics|5": {
"acc": 0.23529411764705882,
"acc_stderr": 0.04220773659171453,
"acc_norm": 0.23529411764705882,
"acc_norm_stderr": 0.04220773659171453
},
"harness|hendrycksTest-computer_security|5": {
"acc": 0.23,
"acc_stderr": 0.04229525846816505,
"acc_norm": 0.23,
"acc_norm_stderr": 0.04229525846816505
},
"harness|hendrycksTest-conceptual_physics|5": {
"acc": 0.2680851063829787,
"acc_stderr": 0.028957342788342347,
"acc_norm": 0.2680851063829787,
"acc_norm_stderr": 0.028957342788342347
},
"harness|hendrycksTest-econometrics|5": {
"acc": 0.24561403508771928,
"acc_stderr": 0.040493392977481404,
"acc_norm": 0.24561403508771928,
"acc_norm_stderr": 0.040493392977481404
},
"harness|hendrycksTest-electrical_engineering|5": {
"acc": 0.2482758620689655,
"acc_stderr": 0.036001056927277716,
"acc_norm": 0.2482758620689655,
"acc_norm_stderr": 0.036001056927277716
},
"harness|hendrycksTest-elementary_mathematics|5": {
"acc": 0.24074074074074073,
"acc_stderr": 0.0220190800122179,
"acc_norm": 0.24074074074074073,
"acc_norm_stderr": 0.0220190800122179
},
"harness|hendrycksTest-formal_logic|5": {
"acc": 0.23015873015873015,
"acc_stderr": 0.03764950879790605,
"acc_norm": 0.23015873015873015,
"acc_norm_stderr": 0.03764950879790605
},
"harness|hendrycksTest-global_facts|5": {
"acc": 0.18,
"acc_stderr": 0.038612291966536934,
"acc_norm": 0.18,
"acc_norm_stderr": 0.038612291966536934
},
"harness|hendrycksTest-high_school_biology|5": {
"acc": 0.25483870967741934,
"acc_stderr": 0.024790118459332208,
"acc_norm": 0.25483870967741934,
"acc_norm_stderr": 0.024790118459332208
},
"harness|hendrycksTest-high_school_chemistry|5": {
"acc": 0.19704433497536947,
"acc_stderr": 0.02798672466673622,
"acc_norm": 0.19704433497536947,
"acc_norm_stderr": 0.02798672466673622
},
"harness|hendrycksTest-high_school_computer_science|5": {
"acc": 0.22,
"acc_stderr": 0.041633319989322695,
"acc_norm": 0.22,
"acc_norm_stderr": 0.041633319989322695
},
"harness|hendrycksTest-high_school_european_history|5": {
"acc": 0.19393939393939394,
"acc_stderr": 0.0308741451365621,
"acc_norm": 0.19393939393939394,
"acc_norm_stderr": 0.0308741451365621
},
"harness|hendrycksTest-high_school_geography|5": {
"acc": 0.3484848484848485,
"acc_stderr": 0.033948539651564025,
"acc_norm": 0.3484848484848485,
"acc_norm_stderr": 0.033948539651564025
},
"harness|hendrycksTest-high_school_government_and_politics|5": {
"acc": 0.32124352331606215,
"acc_stderr": 0.033699508685490674,
"acc_norm": 0.32124352331606215,
"acc_norm_stderr": 0.033699508685490674
},
"harness|hendrycksTest-high_school_macroeconomics|5": {
"acc": 0.23333333333333334,
"acc_stderr": 0.021444547301560476,
"acc_norm": 0.23333333333333334,
"acc_norm_stderr": 0.021444547301560476
},
"harness|hendrycksTest-high_school_mathematics|5": {
"acc": 0.2851851851851852,
"acc_stderr": 0.027528599210340492,
"acc_norm": 0.2851851851851852,
"acc_norm_stderr": 0.027528599210340492
},
"harness|hendrycksTest-high_school_microeconomics|5": {
"acc": 0.29831932773109243,
"acc_stderr": 0.029719142876342856,
"acc_norm": 0.29831932773109243,
"acc_norm_stderr": 0.029719142876342856
},
"harness|hendrycksTest-high_school_physics|5": {
"acc": 0.2781456953642384,
"acc_stderr": 0.03658603262763744,
"acc_norm": 0.2781456953642384,
"acc_norm_stderr": 0.03658603262763744
},
"harness|hendrycksTest-high_school_psychology|5": {
"acc": 0.26788990825688075,
"acc_stderr": 0.018987462257978652,
"acc_norm": 0.26788990825688075,
"acc_norm_stderr": 0.018987462257978652
},
"harness|hendrycksTest-high_school_statistics|5": {
"acc": 0.4351851851851852,
"acc_stderr": 0.03381200005643525,
"acc_norm": 0.4351851851851852,
"acc_norm_stderr": 0.03381200005643525
},
"harness|hendrycksTest-high_school_us_history|5": {
"acc": 0.2647058823529412,
"acc_stderr": 0.0309645179269234,
"acc_norm": 0.2647058823529412,
"acc_norm_stderr": 0.0309645179269234
},
"harness|hendrycksTest-high_school_world_history|5": {
"acc": 0.28270042194092826,
"acc_stderr": 0.029312814153955927,
"acc_norm": 0.28270042194092826,
"acc_norm_stderr": 0.029312814153955927
},
"harness|hendrycksTest-human_aging|5": {
"acc": 0.31390134529147984,
"acc_stderr": 0.031146796482972465,
"acc_norm": 0.31390134529147984,
"acc_norm_stderr": 0.031146796482972465
},
"harness|hendrycksTest-human_sexuality|5": {
"acc": 0.2595419847328244,
"acc_stderr": 0.03844876139785271,
"acc_norm": 0.2595419847328244,
"acc_norm_stderr": 0.03844876139785271
},
"harness|hendrycksTest-international_law|5": {
"acc": 0.2231404958677686,
"acc_stderr": 0.03800754475228733,
"acc_norm": 0.2231404958677686,
"acc_norm_stderr": 0.03800754475228733
},
"harness|hendrycksTest-jurisprudence|5": {
"acc": 0.25925925925925924,
"acc_stderr": 0.042365112580946336,
"acc_norm": 0.25925925925925924,
"acc_norm_stderr": 0.042365112580946336
},
"harness|hendrycksTest-logical_fallacies|5": {
"acc": 0.25153374233128833,
"acc_stderr": 0.03408997886857529,
"acc_norm": 0.25153374233128833,
"acc_norm_stderr": 0.03408997886857529
},
"harness|hendrycksTest-machine_learning|5": {
"acc": 0.29464285714285715,
"acc_stderr": 0.043270409325787296,
"acc_norm": 0.29464285714285715,
"acc_norm_stderr": 0.043270409325787296
},
"harness|hendrycksTest-management|5": {
"acc": 0.17475728155339806,
"acc_stderr": 0.037601780060266224,
"acc_norm": 0.17475728155339806,
"acc_norm_stderr": 0.037601780060266224
},
"harness|hendrycksTest-marketing|5": {
"acc": 0.20085470085470086,
"acc_stderr": 0.026246772946890488,
"acc_norm": 0.20085470085470086,
"acc_norm_stderr": 0.026246772946890488
},
"harness|hendrycksTest-medical_genetics|5": {
"acc": 0.3,
"acc_stderr": 0.046056618647183814,
"acc_norm": 0.3,
"acc_norm_stderr": 0.046056618647183814
},
"harness|hendrycksTest-miscellaneous|5": {
"acc": 0.23499361430395913,
"acc_stderr": 0.01516202415227844,
"acc_norm": 0.23499361430395913,
"acc_norm_stderr": 0.01516202415227844
},
"harness|hendrycksTest-moral_disputes|5": {
"acc": 0.23699421965317918,
"acc_stderr": 0.02289408248992599,
"acc_norm": 0.23699421965317918,
"acc_norm_stderr": 0.02289408248992599
},
"harness|hendrycksTest-moral_scenarios|5": {
"acc": 0.23798882681564246,
"acc_stderr": 0.014242630070574915,
"acc_norm": 0.23798882681564246,
"acc_norm_stderr": 0.014242630070574915
},
"harness|hendrycksTest-nutrition|5": {
"acc": 0.23202614379084968,
"acc_stderr": 0.024170840879341005,
"acc_norm": 0.23202614379084968,
"acc_norm_stderr": 0.024170840879341005
},
"harness|hendrycksTest-philosophy|5": {
"acc": 0.1864951768488746,
"acc_stderr": 0.02212243977248077,
"acc_norm": 0.1864951768488746,
"acc_norm_stderr": 0.02212243977248077
},
"harness|hendrycksTest-prehistory|5": {
"acc": 0.24074074074074073,
"acc_stderr": 0.02378858355165854,
"acc_norm": 0.24074074074074073,
"acc_norm_stderr": 0.02378858355165854
},
"harness|hendrycksTest-professional_accounting|5": {
"acc": 0.2695035460992908,
"acc_stderr": 0.026469036818590627,
"acc_norm": 0.2695035460992908,
"acc_norm_stderr": 0.026469036818590627
},
"harness|hendrycksTest-professional_law|5": {
"acc": 0.2529335071707953,
"acc_stderr": 0.011102268713839989,
"acc_norm": 0.2529335071707953,
"acc_norm_stderr": 0.011102268713839989
},
"harness|hendrycksTest-professional_medicine|5": {
"acc": 0.4411764705882353,
"acc_stderr": 0.030161911930767102,
"acc_norm": 0.4411764705882353,
"acc_norm_stderr": 0.030161911930767102
},
"harness|hendrycksTest-professional_psychology|5": {
"acc": 0.25,
"acc_stderr": 0.01751781884501444,
"acc_norm": 0.25,
"acc_norm_stderr": 0.01751781884501444
},
"harness|hendrycksTest-public_relations|5": {
"acc": 0.21818181818181817,
"acc_stderr": 0.03955932861795833,
"acc_norm": 0.21818181818181817,
"acc_norm_stderr": 0.03955932861795833
},
"harness|hendrycksTest-security_studies|5": {
"acc": 0.20408163265306123,
"acc_stderr": 0.025801283475090506,
"acc_norm": 0.20408163265306123,
"acc_norm_stderr": 0.025801283475090506
},
"harness|hendrycksTest-sociology|5": {
"acc": 0.24378109452736318,
"acc_stderr": 0.03036049015401465,
"acc_norm": 0.24378109452736318,
"acc_norm_stderr": 0.03036049015401465
},
"harness|hendrycksTest-us_foreign_policy|5": {
"acc": 0.24,
"acc_stderr": 0.04292346959909281,
"acc_norm": 0.24,
"acc_norm_stderr": 0.04292346959909281
},
"harness|hendrycksTest-virology|5": {
"acc": 0.22289156626506024,
"acc_stderr": 0.03240004825594687,
"acc_norm": 0.22289156626506024,
"acc_norm_stderr": 0.03240004825594687
},
"harness|hendrycksTest-world_religions|5": {
"acc": 0.3216374269005848,
"acc_stderr": 0.03582529442573122,
"acc_norm": 0.3216374269005848,
"acc_norm_stderr": 0.03582529442573122
},
"harness|truthfulqa:mc|0": {
"mc1": 0.24112607099143207,
"mc1_stderr": 0.014974827279752334,
"mc2": 0.3762297840067963,
"mc2_stderr": 0.01445991036363257
},
"harness|winogrande|5": {
"acc": 0.5153906866614049,
"acc_stderr": 0.014045826789783668
},
"harness|gsm8k|5": {
"acc": 0.006823351023502654,
"acc_stderr": 0.0022675371022544823
}
}
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 28.03 |
| AI2 Reasoning Challenge (25-Shot) | 22.87 |
| HellaSwag (10-Shot) | 30.41 |
| MMLU (5-Shot) | 25.06 |
| TruthfulQA (0-shot) | 37.62 |
| Winogrande (5-shot) | 51.54 |
| GSM8k (5-shot) | 0.68 |