Alternatives
Products that do what Distilled 0.6B text-to-SQL model does
We used our platform to fine-tune a tiny text-to-SQL model using distillation from DeepSeek V3. Repo has instructions for how to replicate this. This is definitely not the best-performing model like this out there! But I found it surprising we were able to get to this much out of it: stone's throw away from a teacher 1000x the size! We also ran the same thing using the 4B Qwen and matched the teacher accuracy, though here the difference is merely 100x :) I find this pretty cool - obviously our distilled models can only do this one task and don't generalize, but that's often exactly what you…
- 1
- 2
- 3DS
I played around with GPT-3 to build this demo. Select a public BigQuery dataset and describe your query in natural English, then edit the generated SQL as needed and execute it. https://app.tabbydata.com/sql-assistant-demo
2021
- 4
- 5RC
2020 · github.com
- 6PA
2021 · github.com
- 7SA
2021 · github.com
- 8FS
A minimalistic Fluent SQL API for Java built with one compilation unit aimed to resemble the code to your original SQL code
2013 · github.com
- 9MM
Hi HN! We (Thomas and Stéphan, hello!) recently released Model2Vec, a Python library for distilling any sentence transformer into a small set of static embeddings. This makes inference with such a model up to 500x faster, and reduces model size by a factor of 15 (7.5M params or 15/30MB on disk, depending on whether you use float16 or float32). This allows you to embed 50-100k documents per second on a cpu on a macbook. This reduction of course comes at a cost: distilled models are worse than their parent models. Even so, they are actually a lot better than large sets of conventional…
2024 · github.com
- 10PP
Hello HN! We’ve been working hard on Vanna, our RAG framework for SQL generation and we’ve been updating our documentation. Please have a look — we have a ton of Jupyter notebooks for any combination of desired use cases. At it’s heart, we have abstractions that help you: - “train” a RAG “model” i.e. add metadata for the retrieval augmentation system to reference when constructing the LLM prompt (yes, we know that the terms “train” and “model” are somewhat confusing and we’re open to changing those terms if you can suggest better ones) - “ask” questions, which will generate SQL, run it,…
2023 · github.com
- 11TA
Hello HN! I'm Peter, one of the folks who helped create Tidepool. We last shared Tidepool with HN about 7 months ago https://news.ycombinator.com/item?id=36957762 Since then, the AI field has moved incredibly quickly and we’ve iterated a lot on our product! The core problem we are trying to solve is: there's a lot of useful business insights you can get from text data, but it's hard to do analytics on it. - SQL is built for tabular / structured data, but when it comes to text, the best you can do is do keyword search. - In the pre-LLM world, you might resort to training a…
2024 · tidepool.so
- 12MM
Hi HN! We (Thomas and Stéphan, hello!) recently released Model2Vec, a Python library for distilling any sentence transformer into a small set of static embeddings. This makes inference with such a model up to 500x faster, and reduces model size by a factor of 15 (7.5M params or 15/30MB on disk, depending on whether you use float16 or float32). This reduction of course comes at a cost: distilled models are a lot worse than their parent models. Even so, they are actually a lot better than large sets of conventional static embeddings, such as GLoVe or word2vec-based models, which are many…
2024 · github.com
- 13LS
Hi HN, I built llm.sql, an LLM inference framework that reimagines the LLM execution pipeline as a series of structured SQL queries atop SQLite. The motivation: Edge LLMs are getting better, but hardware remains a bottleneck, especially RAM (size and bandwidth). When available memory is less than the model size and KV cache, the OS incurs page faults and swaps pages using LRU-like strategies, resulting in throughput degradation that's hard to notice and even harder to debug. In fact, the memory access pattern during LLM inference is deterministic - we know exactly which weights are needed…
Apr 2026
- 14PA
I'm excited to share "take 2" of the Prela query language. After sharing the previous version here, I've received some valuable feedback, the main one being the weird unicode-based syntax throwing people off. Prela now has a more familiar SQL-like syntax while adhering to the algebraic principle, which makes the language compositional and controllable, all the while keeping the core engine under 1k lines of code. The engine has also been rewritten from Julia to Rust, resulting in both simpler code and faster performance (not just because "Rust fast Julia slow", but for some pretty deep…
Jun 2026 · prela-lang.org
- 15WM
Hey HN — We're excited to share Trellis — a snowflake for unstructured data. We've built an AI engine that turns unstructured data into structured SQL-format based on the schema you define in natural language. We spent a lot of time building ML infrastructure and realized that most data warehouses and data pipelines are not designed for unstructured data (documents, PDFs, calls). While something like a Vector database and RAG are great at search tasks, they really struggle with aggregation and SQL type queries such as 1. How many emails in the past 6 months contain complaints about the…
2024 · demo.runtrellis.com
- 16AP
Hello Hacker News, I’m releasing TXT Blah Blah Blah Lite, an open-source plain-text AI reasoning engine powered by semantic embedding rotation. It generates 50 coherent, self-consistent answers within 60 seconds — no training, no external APIs, and zero network calls. Why this matters Six top AI models (ChatGPT, Grok, DeepSeek, Gemini, Perplexity, Kimi) independently gave it perfect 100/100 ratings. For context: Grok scores LangChain around 90 MemoryGPT scores about 92 Typical open-source LLM frameworks score 80-90 Key features Lightweight and portable: runs fully offline as a single…
2025 · github.com
- 17AW
I've spent the past couple of weeks building a Wasm-to-Go translator. It supports a subset of Wasm useful enough to translate SQLite into 600k LoC (~20 MiB) of Go code. It already passes all of my Go SQLite driver's tests across the 20 platforms I support. Performance compared to https://wazero.io/ is a bit of a mixed bag: code that frequently crosses the Go-Wasm boundary improves, but code that spends most of its time in "Wasm land" doesn't. There's probably room for improvement (I'd love to hear your ideas), but this is also a testament to how good the wazero AOT compiler…
Feb 2026 · github.com
- 18ML
2017 · migra.readthedocs.io
- 19
- 20UA
Hey HN! After using a combination of Unsloth and Axolotl a lot, and finding it generally painful to figure out the right performance tuning for things like batch sizing and multi-GPU sharding, I wrote a small Python lib that sets up known-good LoRA training configurations for Llama 3.1 8B and 70B Instruct, and includes helpers for distilling from larger models or training on serverless finetuning platforms, and includes a walkthrough for distilling DeepSeek-R1 into a Llama 3.1 8B LoRA... But you can use it for pretty much any finetuning task, not just distilling large models!
2025 · github.com
- 21LR
I just noticed it takes literally ~5 minutes to train millions parameters on slow CPU...but before you call Yudkowsky that "it's over", an important note: the main bottleneck is the corpus size, params are just 'cleverness' but given limited info it's powerless. Anyway, here is the project: https://github.com/bggb7781-collab/lrnnsmdds/tree/main couple of notes: 1. single C file, no dependencies. Below are literally all the "dependencies", not even custom header (copy paste from the top of the single c file): #define _POSIX_C_SOURCE 200809L #include #include…
Apr 2026 · raw.githubusercontent.com
- 22QP
2020 · github.com
- 23PR
Hi HN, While building RAG agents, I noticed a lot of token budget was wasted on formatting overhead (HTML tags, JSON structure, whitespace). Existing solutions felt too heavy (often requiring torch/transformers), so I wrote this lightweight, zero-dependency library to solve it. It includes strategies for context packing, PII redaction, and tool output compression. Benchmarks show it can save ~15% of tokens with negligible latency overhead (<0.5ms). Happy to answer any questions!
Dec 2025 · github.com
- 24OD
I launched my startup's Snowflake optimization product here about two years ago, and we've just released our second product: a Databricks SQL optimizer. The optimizations are basically kubernetes for data warehousing: we take over autoscaling and cluster selection to increase utilization without impacting latency. The scheduler is backed by ML models that predict runtime and capacity, which means we can run machines hotter than the providers can and thereby cut costs. More info here: https://espresso.ai/post/launching-our-databricks-sql-optimi...
Oct 2025
Ranked by how close each launch is in meaning, then by votes. Refine with a description →