gauntlet results

These are are this model's output results on my "summarization gauntlet". You can find more info about that here on my dropbox for it or at this dataset.

if you aren't familiar with it, one thing to note is some of the docs purposefully are "messy"/have spelling errors etc.

parameters

{
  "model_name_or_path": "pszemraj/pegasus-x-large-book_synthsumm",
  "use_cuda": true,
  "token_batch_length": 16384,
  "batch_stride": 16,
  "max_length_ratio": 0.25,
  "load_in_8bit": false,
  "compile_model": true,
  "optimum_onnx": false,
  "device": "cuda",
  "inference_params": {
    "min_length": 8,
    "max_length": 4096,
    "no_repeat_ngram_size": 3,
    "encoder_no_repeat_ngram_size": 4,
    "repetition_penalty": 2.5,
    "num_beams": 10,
    "num_beam_groups": 1,
    "length_penalty": 1.0,
    "early_stopping": true,
    "do_sample": false
  },
  "textsum_version": "0.2.0"
}

Created: 2023-11-28T19:05:49.579673

ASR-whisper-rpunctuated_Noam Chomsky, Fundam_1669853561_0_part1_summary

The speaker discusses fundamental computational operations in constructing syntactic objects, emphasizing the importance of genuine explanations and reorganization of data. They also discuss the limitations of neural nets for computation and the role of the brain in language development.

Section Scores for ASR-whisper-rpunctuated_Noam Chomsky, Fundam_1669853561_0_part1_summary:

-0.6261

ASR-whisper-rpunctuated_Noam Chomsky, Fundam_1669853631_0_part2_summary

The speaker discusses the concept of merge in linguistics, focusing on its limitations and potential applications. They emphasize the need for a genuine explanation and explain how it can be used to solve complex problems such as multidimensionality and unbounded coordination. The speaker discusses the use of the pair-merge element to block in situ and raising cases, as well as head movement. They also discuss the limitations of traditional methods for analyzing head movement and the need for a more principled approach to explanation.

Section Scores for ASR-whisper-rpunctuated_Noam Chomsky, Fundam_1669853631_0_part2_summary:

-0.8944
-0.7277

ASRnlp_law_lecture_week_1_v_2_c_transcription_1_summary

The speaker provides an overview of the course "Natural Language Processing for Law & Social Science," covering topics such as dictionary methods, tocanization, and sentiment analysis. They also discuss the application of these methods in legal and social science, emphasizing the importance of understanding language models and their limitations.

Section Scores for ASRnlp_law_lecture_week_1_v_2_c_transcription_1_summary:

-0.7283

ASRnlp_law_lecture_week_2_v_2_c_transcription_2_summary

The speaker updates the class on upcoming assignments, homework emissions, and other topics related to computer science. They also discuss the importance of document representation learning and transforming documents into restrictions, as well as the use of washing vacterizers and feature selection in text classification. Additionally, they provide examples of successful course projects and offer guidance on selecting a topic for future courses.

Section Scores for ASRnlp_law_lecture_week_2_v_2_c_transcription_2_summary:

-0.7209

ASRnlp_law_lecture_week_3_part_1_v_2_c_transcription_3_summary

The speaker discusses the use of feature selection techniques in social science to identify words associated with specific political parties. They also discuss methods for document distance and dimension reduction, as well as their applications in policy analysis and finance.

Section Scores for ASRnlp_law_lecture_week_3_part_1_v_2_c_transcription_3_summary:

-0.7762

Emie_dissertation_cleansed_summary

The dissertation explores the relationship between urban space, material reality, and movement in post-war American and British film noir films. It examines Alfred Zinn's "Act of Violence," a 1948 film noir about a WWII veteran named Frank Enley, who flees Los Angeles to protect his friend Joe Parkson from a German hitman. The film also explores the characters' struggles with reconfiguring their identities through movement and speed, highlighting the tension between realism and formative tendencies in these films. The film "The Man Between," directed by Carol Reed and Alfred Zinnemann, explores the relationship between material reality and character movement in post-war urban environments. It emphasizes the importance of confronting material reality to escape trauma and identity loss. The film also challenges the theory of film as a balm for modern abstraction.

Section Scores for Emie_dissertation_cleansed_summary:

-0.8099
-0.7794

OCR_ML4HLecture02image__summary

Gunnar Ratsch and Julia Vogt presented a lecture on machine learning for medical image analysis, covering topics such as segmentation, superpixels, Markov random fields, image classification, neural networks, and clinical-grade decision support. They also highlighted the potential use of deep learning in medical imaging and proposed innovative ways to integrate computational pathology into the clinical workflow.

Section Scores for OCR_ML4HLecture02image__summary:

-0.4782

OCR_ML4HLecture04RepresentationLearning.pptx__summary

Gunnar Ratsch presented a lecture on machine learning for health care, covering topics such as computational patient representations, autoencoders, sequence to sequence models, ICU benchmarking, generative models, VAEs, contrastive learning, and improving clinical predictions through unsupervised time series representation learning. He also discussed the use of generative models for interpreting health state representations and the limitations of existing EHR datasets.

Section Scores for OCR_ML4HLecture04RepresentationLearning.pptx__summary:

-0.6116

OCR_ML4HLecture05-NLP.pptx__summary

Gunnar Ratsch and Rita Kuznetsova presented a lecture on machine learning for health care, covering topics such as basic preprocessing steps, text features, LDA algorithms, word embedding models, POS tagging, language modeling, and the use of deep learning in NLP. They also discussed the importance of patient models for precision medicine, clinical notes, and PubMed abstracts.

Section Scores for OCR_ML4HLecture05-NLP.pptx__summary:

-0.6503

OCR_PAPER_Hong et al. - 2022 - CogVideo Large-scale Pretraining for Text-to-Video Generation via Transformers-annotated__summary

The paper explores the use of large-scale transformers for video generation and presents Cog Video, an open-source transformer with 9B parameters trained on 5 million pairs. It introduces a multi-frame rate hierarchical training strategy and demonstrates its effectiveness in generating high-resolution videos. The paper also discusses the human evaluation process and provides examples of samples generated by the model.

Section Scores for OCR_PAPER_Hong et al. - 2022 - CogVideo Large-scale Pretraining for Text-to-Video Generation via Transformers-annotated__summary:

-0.8547

OCR_PAPER_Kandpal, Nieto, Jin - 2022 - Music Enhancement via Image Translation and Vocoding-annotated__summary

The paper explores a solution to enhance music signals through image transformation and waveform synthesis, focusing on polyphonic signal enhancement. It compares the effectiveness of this approach with classical methods and uses a listening test to evaluate its reliability. The authors hope their work will inspire further research and promote the field of music enhancement.

Section Scores for OCR_PAPER_Kandpal, Nieto, Jin - 2022 - Music Enhancement via Image Translation and Vocoding-annotated__summary:

-0.749

OCR_PAPER_dall-e-2-annotated__summary

The paper explores the use of Contrastive Learning (CLIP) for image generation using diffusion models and computer guidance techniques. It shows that explicitly generating images improves image diversity, preserves semantic information, and enables language-guided manipulations. The authors compare their samples with other systems and find that diffusion models outperform autoregressive ones in terms of sample quality and computational efficiency. They also conduct human evaluations to assess the effectiveness of their system.

Section Scores for OCR_PAPER_dall-e-2-annotated__summary:

-0.8006

The Most Dangerous Game--Richard Connell_summary

Richard Connell's "The Most Dangerous Game" is about a sailor named Rainsford who encounters a giant hunter named General Zaroff on the island of Ship-Trap Island. The two men engage in a game of hunting, which turns out to be one of the most dangerous and thrilling hunts in the book. After a year of survival efforts, they capture the giant hunter and win the game.

Section Scores for The Most Dangerous Game--Richard Connell_summary:

-0.9355

gpt_peter_testing_group_exemplars_summary

The text is a collection of conversations and messages between various individuals discussing topics such as Korea, consciousness, self-awareness, cultural differences, hobbies, relationships, hackathons, Star Wars 11, sustainability, quantum machine learning, human manipulation, and personal growth. It also includes hints of potential future collaborations and interactions with Amazon support representatives.

Section Scores for gpt_peter_testing_group_exemplars_summary:

-0.8382

navy seals copy pasta_summary

The main character is a sniper with extensive combat experience and access to the arsenal of the United States Marine Corps. He plans to wipe out his enemies with precision, using unarmed combat and his network of spies in the U.S. to do so.

Section Scores for navy seals copy pasta_summary:

-0.8124

script_findingnemo_summary

"Finding Nemo" is a Disney film about a clownfish named Nemo and his father, Marlin. Nemo's first day at school is interrupted by a search for his son, who has been taken away by divers. Despite the efforts of various characters, Nemo manages to escape and reunite with his family in Sydney. The film "Finding Nemo" follows the adventures of Dory, a sea turtle, and her mother, Marlin, as they struggle to find their son Nemo after being stranded in a whale's mouth. Finding Nemo is a Disney film directed by Andrew Stanton and written by Bob Peterson & David Reynolds.

Section Scores for script_findingnemo_summary:

-0.7657
-0.7198

script_frozendisney_summary

"Frozen" is a Disney film directed by Jennifer Lee about a princess named Anna who falls in love with a man named Hans. Despite her family's protests, she marries him and becomes the new queen of the kingdom, despite Elsa's icy powers. The film explores themes of love, friendship, and the challenges of being a princess in a frozen world. The film "Frozen" follows the adventures of Anna, Olaf, and Kristoff as they navigate through icy terrain to rescue Anna from Elsa's ice palace. After a series of mishaps, Anna manages to thaw her frozen heart and win Olaf's love. However, she is killed by Elsa in a storm, leading to Frozen's transformation into an ice-skating rink.

Section Scores for script_frozendisney_summary:

-0.8674
-0.8671

script_strangersonatrain_summary

"Strawberries on a Train" is a 1956 film directed by Raymond Chandler about a young man named Bruno Anthony who plans to murder his estranged wife Miriam. Guy Haines, a well-known tennis player, befriends Bruno and becomes involved in the plot. After Miriam's murder, Guy is pursued by the police, who discover that he had an alibi for the murder. The film ends with Anne and Guy embracing and expressing their love for each other. The film "Bruno" follows a tennis player named Guy Haines, who is accused of murdering his wife Miriam and tries to cover it up by playing tennis at the Forest Hills Tennis Club. However, he accidentally drops his cigarette lighter in a public place and is arrested for the murder. The film "Bruno" follows Bruno and Guy Haines as they engage in a tennis match at an amusement park, where Bruno tries to steal Guy's cigarette lighter. Guy is pursued by the police, who discover Bruno's involvement in the murder of his wife. After Bruno dies, Guy offers to stay in town for the night, and Anne receives a phone call from her father.

Section Scores for script_strangersonatrain_summary:

-0.786
-0.8048
-0.8261

script_sunsetblvd._summary

"Sunset Boulevard" is a screenplay by Charles Brackett and Billy Wilder, set in Los Angeles in the early 1900s. It follows the story of Joe Gillis, an opium smuggler, and his relationship with Norma Desmond, a well-known movie actress. The screenplay explores themes of love, betrayal, and the complexities of human relationships. The text is a collection of dialogue and scenes from the film "The Mysteries of Norma Desmond," written by Gordon Cole and directed by Joe Gillis. It explores the complex relationship between Norma, an aging actress, and Gillis, a young director who has fallen in love with her while working on a script. After a series of confrontations and dramatic events, Norma's story ends with her triumphant return to the studio after twenty years.

Section Scores for script_sunsetblvd._summary:

-0.7195
-0.8745