Microsoft says GPT 3.5 has 20B parameters? (arxiv.org)

Screenshot of arxiv.org: Microsoft says GPT 3.5 has 20B parameters?

First screen, captured 3 years ago.

Looks similar

Screenshot of arxiv.org: Neural Network Diffusion

Neural Network Diffusion (arxiv.org)

199 points by vagabund 2 years ago | 74 comments

Screenshot of arxiv.org: TiDAR: Think in Diffusion, Talk in Autoregression

TiDAR: Think in Diffusion, Talk in Autoregression (arxiv.org)

111 points by internetguy 11 months ago | 18 comments

Screenshot of arxiv.org: Type-constrained code generation with language models

Type-constrained code generation with language models (arxiv.org)

237 points by tough 1 year ago | 109 comments

Screenshot of arxiv.org: Mercury: Ultra-fast language models based on diffusion

Mercury: Ultra-fast language models based on diffusion (arxiv.org)

557 points by PaulHoule 1 year ago | 230 comments

arXiv paper from Inception Labs presenting Mercury, diffusion-based large language models for coding that generate multiple tokens in parallel at very high throughput.

diffusion-language-models mercury-coder inception-labs llm-inference

Screenshot of arxiv.org: Language models as compilers: Simulating pseudocode execution

Language models as compilers: Simulating pseudocode execution (arxiv.org)

159 points by milliondreams 2 years ago | 49 comments

Screenshot of arxiv.org: Some critical issues with the SWE-bench dataset

Some critical issues with the SWE-bench dataset (arxiv.org)

331 points by joshwa 1 year ago | 113 comments

arXiv paper introducing SWE-Bench+, an empirical analysis revealing data quality issues in the SWE-bench coding benchmark for LLMs, such as solution leakage and weak tests.

swe-bench llm-benchmarks code-generation gpt-4

Looks similar

Neural Network Diffusion(arxiv.org)

TiDAR: Think in Diffusion, Talk in Autoregression(arxiv.org)

Type-constrained code generation with language models(arxiv.org)

Mercury: Ultra-fast language models based on diffusion(arxiv.org)

Language models as compilers: Simulating pseudocode execution(arxiv.org)

Some critical issues with the SWE-bench dataset(arxiv.org)