Hallo everyone!
I will be moving to Essen around mid-September or October and I am currently exploring PhD opportunities at Heinrich Heine University Düsseldorf. I will be living about 20 minutes on foot, or 10 minutes by metro, from Essen Hbf. I was wondering whether commuting daily from Essen to Düsseldorf is realistic, especially by train. Do you have any experience?
Danke schön!
Hello everyone,
Is it allowed to use OpenAI API outputs to create a silver code dataset or benchmark for a specific Python library (EPyT)?
I am working on a project idea related to library-specific code generation. The concrete case is a specific Python library used in a technical/scientific domain. The goal would be to improve and evaluate how well code-generation models can use this library correctly.
I am trying to understand the legal / Terms of Service boundary around using OpenAI API outputs in two different scenarios:
Scenario 1: Silver dataset for fine-tuning an OSS model
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests for the specific Python library.
Then human-review, filter, and validate the generated examples. Then use this silver dataset to fine-tune an open-source code model, with the goal of improving its performance on this specific library.
My question: would this violate OpenAI’s terms because the API outputs are being used to train/fine-tune another coding model, even if the scope is narrow and library-specific?
Scenario 2: Benchmark only, not training
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests.
Human-review and validate them. Then use the resulting dataset only as an evaluation benchmark to compare different models. The benchmark would not be used to fine-tune or train any model.
My question: is this generally considered allowed under OpenAI’s terms, assuming the benchmark is properly reviewed and documented as AI-assisted?
I understand that Reddit is not legal advice, and I would still contact OpenAI or legal counsel for a definitive answer. However, I thought new ideas could come up from people who have already faced similar situations in practice.
Hello everyone,
Is it allowed to use OpenAI API outputs to create a silver code dataset or benchmark for a specific Python library?
I am working on a project idea related to library-specific code generation. The concrete case is a specific Python library used in a technical/scientific domain. The goal would be to improve and evaluate how well code-generation models can use this library correctly.
I am trying to understand the legal / Terms of Service boundary around using OpenAI API outputs in two different scenarios:
Scenario 1: Silver dataset for fine-tuning an OSS model
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests for the specific Python library.
Then human-review, filter, and validate the generated examples. Then use this silver dataset to fine-tune an open-source code model, with the goal of improving its performance on this specific library.
My question: would this violate OpenAI’s terms because the API outputs are being used to train/fine-tune another coding model, even if the scope is narrow and library-specific?
Scenario 2: Benchmark only, not training
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests.
Human-review and validate them. Then use the resulting dataset only as an evaluation benchmark to compare different models. The benchmark would not be used to fine-tune or train any model.
My question: is this generally considered allowed under OpenAI’s terms, assuming the benchmark is properly reviewed and documented as AI-assisted?
I understand that Reddit is not legal advice, and I would still contact OpenAI or legal counsel for a definitive answer. However, I thought new ideas could come up from people who have already faced similar situations in practice.
Thank you in advance!
Hello everyone,
Is it allowed to use OpenAI API outputs to create a silver code dataset or benchmark for a specific Python library (EPyT)?
I am working on a project idea related to library-specific code generation. The concrete case is a specific Python library used in a technical/scientific domain. The goal would be to improve and evaluate how well code-generation models can use this library correctly.
I am trying to understand the legal / Terms of Service boundary around using OpenAI API outputs in two different scenarios:
Scenario 1: Silver dataset for fine-tuning an OSS model
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests for the specific Python library.
Then human-review, filter, and validate the generated examples. Then use this silver dataset to fine-tune an open-source code model, with the goal of improving its performance on this specific library.
My question: would this violate OpenAI’s terms because the API outputs are being used to train/fine-tune another coding model, even if the scope is narrow and library-specific?
Scenario 2: Benchmark only, not training
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests.
Human-review and validate them. Then use the resulting dataset only as an evaluation benchmark to compare different models. The benchmark would not be used to fine-tune or train any model.
My question: is this generally considered allowed under OpenAI’s terms, assuming the benchmark is properly reviewed and documented as AI-assisted?
I understand that Reddit is not legal advice, and I would still contact OpenAI or legal counsel for a definitive answer. However, I thought new ideas could come up from people who have already faced similar situations in practice.
Thank you in advance!
Hello everyone,
Is it allowed to use OpenAI API outputs to create a silver code dataset or benchmark for a specific Python library?
I am working on a project idea related to library-specific code generation. The concrete case is a specific Python library used in a technical/scientific domain. The goal would be to improve and evaluate how well code-generation models can use this library correctly.
I am trying to understand the legal / Terms of Service boundary around using OpenAI API outputs in two different scenarios:
Scenario 1: Silver dataset for fine-tuning an OSS model
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests for the specific Python library.
Then human-review, filter, and validate the generated examples. Then use this silver dataset to fine-tune an open-source code model, with the goal of improving its performance on this specific library.
My question: would this violate OpenAI’s terms because the API outputs are being used to train/fine-tune another coding model, even if the scope is narrow and library-specific?
Scenario 2: Benchmark only, not training
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests.
Human-review and validate them. Then use the resulting dataset only as an evaluation benchmark to compare different models. The benchmark would not be used to fine-tune or train any model.
My question: is this generally considered allowed under OpenAI’s terms, assuming the benchmark is properly reviewed and documented as AI-assisted?
I understand that Reddit is not legal advice, and I would still contact OpenAI or legal counsel for a definitive answer. However, I thought new ideas could come up from people who have already faced similar situations in practice.
Hello everyone,
Is it allowed to use OpenAI API outputs to create a silver code dataset or benchmark for a specific Python library?
I am working on a project idea related to library-specific code generation. The concrete case is a specific Python library used in a technical/scientific domain. The goal would be to improve and evaluate how well code-generation models can use this library correctly.
I am trying to understand the legal / Terms of Service boundary around using OpenAI API outputs in two different scenarios:
Scenario 1: Silver dataset for fine-tuning an OSS model
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests for the specific Python library.
Then human-review, filter, and validate the generated examples. Then use this silver dataset to fine-tune an open-source code model, with the goal of improving its performance on this specific library.
My question: would this violate OpenAI’s terms because the API outputs are being used to train/fine-tune another coding model, even if the scope is narrow and library-specific?
Scenario 2: Benchmark only, not training
Use the OpenAI API to generate programming tasks, reference solutions, and verification tests.
Human-review and validate them. Then use the resulting dataset only as an evaluation benchmark to compare different models. The benchmark would not be used to fine-tune or train any model.
My question: is this generally considered allowed under OpenAI’s terms, assuming the benchmark is properly reviewed and documented as AI-assisted?
I understand that Reddit is not legal advice, and I would still contact OpenAI or legal counsel for a definitive answer. However, I thought new ideas could come up from people who have already faced similar situations in practice.