Evaluating LLMs on creative writing via reader usage, not benchmarks
Hey HN! I'd love to get some people to mess around with a little side project I built to teach myself DSPy! I've been a big fan of reading fiction + webnovels for a while now, and have always been curious about two things: how can LLMs iteratively learn to write better based on reader feedback, and which LLMs are actually best at creative writing (research benchmarks are cool, but don't necessarily translate to real-world usage). That's exactly why I built narrator.sh! The platform takes in a user input for a novel idea, then generates serialized fiction chapter-by-chapter by using DSPy to…
Evaluating LLMs on creative writing via reader usage, not benchmarks launched on August 14, 2025 with 36 votes, #520 of 8,530 launches that month and more than 90% of that year's launches.
In plain words
Narrator.sh is a platform that generates serialized fiction chapter-by-chapter based on user story ideas. It uses DSPy to optimize LLM writing output by incorporating real reader feedback across chapters, employing chain-of-thought reasoning and reward functions to improve subsequent writing. The tool is designed for readers and writers interested in exploring which LLMs perform best at creative writing in practical settings, moving beyond traditional research benchmarks to evaluate performance through actual usage and reader preferences.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hey HN! I'd love to get some people to mess around with a little side project I built to teach myself DSPy! I've been a big fan of reading fiction + webnovels for a while now, and have always been curious about two things: how can LLMs iteratively learn to write better based on reader feedback, and which LLMs are…
Does a similar job
all alternatives →
I used Claude Code to discover connections between 100 booksJan 2026 · trails.pieterma.es · ▲524I think LLMs are overused to summarise and underused to help us read deeper. I built a system for Claude Code to browse 100 non-fiction books and find interesting connections between them. I started out with a pipeline in stages, chaining together LLM calls to build up a context of the library. I was mainly getting back the insight that I was baking into the prompts, and the results weren't particularly surprising. On a whim, I gave CC access to my debug CLI tools and found that it wiped the floor with that approach. It gave actually interesting results and required very little orchestration…
- LSLists.sh – A Microblog for Lists2022 · lists.sh · ▲271
Greetings, creator here! I've been working on a new blogging platform specifically for lists on and off for a few months now and I'm excited to officially announce its launch. After seeing https://charm.sh a few months ago, I've been enamored by the idea of SSH apps. I decided that a blogging platform focused on developers could be the perfect use case for an SSH app. Also, I love writing lists. I think restricting writing to a set of lists can really help improve clarity in thought. The goal of this blogging platform is to make it simple to use the tools you love to write and…
- ISI scraped 3B Goodreads reviews to train a better recommendation modelNov 2025 · book.sv · ▲606
Hi everyone, For the past couple months I've been working on a website with two main features: - https://book.sv - put in a list of books and get recommendations on what to read next from a model trained on over a billion reviews - https://book.sv/intersect - put in a list of books and find the users on Goodreads who have read them all (if you don't want to be included in these results, you can opt-out here: https://book.sv/remove-my-data) Technical info available here: https://book.sv/how-it-works Note 1: If you only provide one or two…
Mapping LLM Style and Range in Flash Fiction2025 · github.com · ▲7Documents the style side of the short-story Creative Writing LLM benchmark: we generated many short stories with a range of LLMs, then analyzed those stories for stylistic fingerprints and within-model diversity. This study focuses on how models write, how their outputs differ, and how varied each m
LLM, a Rust Crate/CLI for CPU Inference of LLMs (LLaMA, GPT-NeoX, etc.)2023 · github.com · ▲45G'day, HN! I'm one of the maintainers of `llm`. I've been working alongside a trusty group of contributors to bring this project to life, and we're now at a point where we're ready to share it with the world. Large language models (LLMs) are taking the computing world by storm due to their emergent abilities that allow them to perform a wide variety of tasks, including translation, summarization, code generation, and even some degree of reasoning. However, the ecosystem around LLMs is still in its infancy, and it can be difficult to get started with these models. `llm` is a one-stop shop for…
A new benchmark for testing LLMs for deterministic outputsApr 2026 · interfaze.ai · ▲60When building workflows that rely on LLMs, we commonly use structured output for programmatic use cases like converting an invoice into rows or meeting transcripts into tickets or even complex PDFs into database entries. The model may return the schema you want, but with hallucinated values like `invoice_date` being off by 2 months or the transcript array ordered wrongly. The JSON is valid, but the values are not. Structured output today is a big part of using LLMs, especially when building deterministic workflows. Current structured output benchmarks (e.g., JSONSchemaBench) only validate…
More work this month
the category →
Endless lofi beats composed live in your browser, over pixel-art nights in Paris, Tokyo, New York, Hong Kong, Sydney, San Francisco and more. Free, no sign-up.
Work · 25d ago · loficities.com
Smooth Recorder▲257Mac screen recordings that look edited when you hit stop
Work · 19d ago · smoothrecorder.com
Work · 13d ago · readrare.com
Source: Show HN post ↗ · launch text quoted in part
Launched alongside, August 2025
the whole month →
I built the world's most impractical 1000-pixel display and anyone in the world can draw on it. It draws a single pixel at a time and takes 30-60 minutes to complete a single image. Anyone can participate in the project by voting for the next image to be drawn, and submitting images. https://kilopx.com/
Work · 2025 · benholmen.com
Kitten TTS is an open-source series of tiny and expressive text-to-speech models for on-device applications. We are excited to launch a preview of our smallest model, which is less than 25 MB. This model has 15M parameters. This release supports English text-to-speech applications in eight voices: four male and four female. The model is quantized to int8 + fp16, and it uses onnx for runtime. The model is designed to run literally anywhere eg. raspberry pi, low-end smartphones, wearables, browsers etc. No GPU required! We're releasing this to give early users a sense of the latency and voices…
AI · 2025 · github.com
I was wondering how I can arrange objects along a spherical helix path, and read some articles on it. I ended up learning about parametric equations again, and make this visualization to document what I learned: https://visualrambling.space/moving-objects-in-3d/ feel free to visit and let me know what you think!
Life & fun · 2025 · visualrambling.space
For HTML Day 2025 [1], I made a web service that displays the current sky at your approximate location as a CSS gradient. Colours are simulated on-demand using atmospheric absorption and scattering coefficients. Updates every minute, without the use of client-side JavaScript. Source code and additional information is available on GitHub: https://github.com/dnlzro/horizon [1] https://html.energy/html-day/2025/index.html
Dev tools · 2025 · sky.dlazaro.ca