README.md

---
datasets:
- tum-nlp/IDMGSP
language:
- en
tags:
- scientific paper
- fake papers
- science
- scientific text
widget:
- text: >
    Abstract:


    The Hartree-Fock (HF) method is a widely used method for approximating the
    electronic structure of many-electron systems. In this work, we study the
    properties of HF solutions of the three-dimensional electron gas (3DEG), a
    model system consisting of a uniform, non-interacting electron gas in three
    dimensions. We find that the HF solutions accurately reproduce the known
    analytic results for the ground state energy and the static structure factor
    of the 3DEG. However, we also find that the HF solutions fail to accurately
    describe the excitation spectrum of the 3DEG, particularly at high energies.


    Introduction:


    The HF method is a self-consistent method for approximating the electronic
    structure of many-electron systems. It is based on the assumption that the
    electrons in a system can be described as non-interacting quasiparticles,
    each with its own effective potential. The HF method is commonly used to
    study the ground state properties of systems, such as the energy and the
    density distribution, but it can also be used to study excited states.


    The 3DEG is a model system that has been widely studied as a test case for
    electronic structure methods. It consists of a uniform, non-interacting
    electron gas in three dimensions, with a finite density and a periodic
    boundary condition. The 3DEG has a number of known analytic results for its
    ground state properties, such as the ground state energy and the static
    structure factor, which can be used to test the accuracy of approximate
    methods.


    Conclusion:


    In this work, we have studied the properties of HF solutions of the 3DEG. We
    find that the HF solutions accurately reproduce the known analytic results
    for the ground state energy and the static structure factor of the 3DEG.
    However, we also find that the HF solutions fail to accurately describe the
    excitation spectrum of the 3DEG, particularly at high energies. This
    suggests that the HF method may not be suitable for accurately describing
    the excited states of the 3DEG. Further work is needed to understand the
    limitations of the HF method and to develop improved methods for studying
    the electronic structure of many-electron systems.
  example_title: Example ChatGPT fake
- text: >
    Abstract:


    Recent calculations have pointed to a 2.8 $\sigma$ tension between data on
    $\epsilon^{\prime}_K / \epsilon_K$ and the standard-model (SM) prediction.
    Several new physics (NP) models can explain this discrepancy, and such NP
    models are likely to predict deviations of $\mathcal{B}(K\to \pi \nu
    \overline{\nu})$ from the SM predictions, which can be probed precisely in
    the near future by NA62 and KOTO experiments. We present correlations
    between $\epsilon^{\prime}_K / \epsilon_K$ and $\mathcal{B}(K\to \pi \nu
    \overline{\nu})$ in two types of NP scenarios: a box dominated scenario and
    a $Z$-penguin dominated one. It is shown that different correlations are
    predicted and the future precision measurements of $K \to \pi \nu
    \overline{\nu}$ can distinguish both scenarios.


    Introduction:


    CP violating flavor-changing neutral current decays of K mesons are extremely
    sensitive to new physics (NP) and can probe virtual effects of particles with
    masses far above the reach of the Large Hadron Collider. Prime examples of
    such observables are ϵ′ K measuring direct CP violation in K → ππ decays and
    B(KL → π0νν). Until recently, large theoretical uncertainties precluded
    reliable predictions for ϵ′ K. Although standard-model (SM) predictions of
    ϵ′ K using chiral perturbation theory are consistent with the experimental
    value, their theoretical uncertainties are large. In contrast, calculation
    by the dual QCD approach 1 finds the SM value much below the experimental
    one. A major breakthrough has been the recent lattice-QCD calculation of the
    hadronic matrix elements by RBC-UKQCD collaboration 2, which gives support
    to the latter result. The SM value at the next-to-leading order divided by
    the indirect CP violating measure ϵK is 3 which is consistent with (ϵ′
    K/ϵK)SM = (1.9±4.5)×10−4 given by Buras et al 4.a Both results are based on
    the lattice numbers, and further use CP-conserving K → ππ data to constrain
    some of the hadronic matrix elements involved. Compared to the world average
    of the experimental results 6, Re (ϵ′ K/ϵK)exp = (16.6 ± 2.3) × 10−4, (2)
    the SM prediction lies below the experimental value by 2.8 σ. Several NP
    models including supersymmetry (SUSY) can explain this discrepancy. It is
    known that such NP models are likely to predict deviations of the kaon rare
    decay branching ratios from the SM predictions, especially B(K → πνν) which
    can be probed precisely in the near future by NA62 and KOTO experiments.b In
    this contribution, we present correlations between ϵ′ K/ϵK and B(K → πνν) in
    two types of NP scenarios: a box dominated scenario and a Z-penguin
    dominated one. Presented at the 52th Rencontres de Moriond electroweak
    interactions and unified theories, La Thuile, Italy, 18-25 March, 2017.
    aOther estimations of the SM value are listed in Kitahara et al 5. b The
    correlations between ϵ′ K/ϵK, B(K → πνν) and ϵK through the CKM components
    in the SM are discussed in Ref. 7.


    Conclusion:


    We have presented the correlations between ϵ′ K/ϵK, B(KL → π0νν), and B(K+ →
    π+νν) in the box dominated scenario and the Z-penguin dominated one. It is
    shown that the constraint from ϵK produces different correlations between two
    NP scenarios. In the future, measurements of B(K → πνν) will be significantly
    improved. The NA62 experiment at CERN measuring B(K+ → π+νν) is aiming to
    reach a precision of 10 % compared to the SM value already in 2018. In order
    to achieve 5% accuracy more time is needed. Concerning KL → π0νν, the KOTO
    experiment at J-PARC aims in a first step at measuring B(KL → π0νν) around
    the SM sensitivity. Furthermore, the KOTO-step2 experiment will aim at 100
    events for the SM branching ratio, implying a precision of 10 % of this
    measurement. Therefore, we conclude that when the ϵ′ K/ϵK discrepancy is
    explained by the NP contribution, NA62 experiment could probe whether a
    modified Z-coupling scenario is realized or not, and KOTO-step2 experiment
    can distinguish the box dominated scenario and the simplified modified
    Z-coupling scenario.
  example_title: Example real
license: openrail++
---
# Model Card for IDMGSP-Galactica-TRAIN

A fine-tuned Galactica model to detect machine-generated scientific papers based on their abstract, introduction, and conclusion. 

This model is trained on the `train` dataset found in https://huggingface.co/datasets/tum-nlp/IDMGSP.

# this model card is WIP, please check the repository, the dataset card and the paper for more details.

## Model Details

### Model Description

<!-- Provide a longer summary of what this model is. -->

- **Developed by:** Technical University of Munich (TUM)
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** English
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** Galactica

### Model Sources

<!-- Provide the basic links for the model. -->

- **Repository:** https://github.com/qwenzo/-IDMGSP
- **Paper:** [More Information Needed]

## Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

### Direct Use

```python
from transformers import AutoTokenizer, OPTForSequenceClassification, pipeline

model = OPTForSequenceClassification.from_pretrained("tum-nlp/IDMGSP-Galactica-TRAIN")
tokenizer = AutoTokenizer.from_pretrained("tum-nlp/IDMGSP-Galactica-TRAIN")
reader = pipeline("text-classification", model=model, tokenizer = tokenizer)
reader(
'''
Abstract:
....

Introduction:
....

Conclusion:
...'''
)
```

### Downstream Use [optional]

<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

[More Information Needed]

### Out-of-Scope Use

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

[More Information Needed]

## Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

[More Information Needed]

## Training Details

### Training Data

The training dataset comprises scientific papers generated by the Galactica, GPT-2, SCIgen, and ChatGPT models, as well as papers extracted from the arXiv database.

The provided table displays the sample counts from each source utilized in constructing the training dataset. 
The dataset could be found in https://huggingface.co/datasets/tum-nlp/IDMGSP.

| Dataset                      | arXiv (real) | ChatGPT (fake) | GPT-2 (fake) | SCIgen (fake) | Galactica (fake) | GPT-3 (fake) |
|------------------------------|--------------|----------------|--------------|----------------|------------------|--------------|
| Standard train (TRAIN)       | 8k           | 2k             | 2k           | 2k             | 2k               | -            |

### Training Procedure 

<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

#### Preprocessing [optional]

[More Information Needed]


#### Training Hyperparameters

[More Information Needed]

#### Speeds, Sizes, Times [optional]

<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->

[More Information Needed]

## Evaluation

<!-- This section describes the evaluation protocols and provides the results. -->

### Testing Data, Factors & Metrics

#### Testing Data

<!-- This should link to a Data Card if possible. -->

[More Information Needed]

#### Factors

<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->

[More Information Needed]

#### Metrics

<!-- These are the evaluation metrics being used, ideally with a description of why. -->

[More Information Needed]

### Results

[More Information Needed]

#### Summary



## Model Examination [optional]

<!-- Relevant interpretability work for the model goes here -->

[More Information Needed]

## Environmental Impact

<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->

- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]

## Technical Specifications [optional]

### Model Architecture and Objective

[More Information Needed]

### Compute Infrastructure

[More Information Needed]

#### Hardware

[More Information Needed]

#### Software

[More Information Needed]

## Citation [optional]

<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->

**BibTeX:**

[More Information Needed]

**APA:**

[More Information Needed]

## Glossary [optional]

<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->

[More Information Needed]

## More Information [optional]

[More Information Needed]

## Model Card Authors [optional]

[More Information Needed]

## Model Card Contact

[More Information Needed]