Timber
AI
classical ml inference
SHOW HN2026
How HN took it
6 positive, 6 neutral or asking, 0 mixed, 5 critical, out of 17 comments about the project. Thread score 0.00.
Idea: 3 for, 2 againstDesign: 2 for, 1 againstPerformance: 1 for, 1 againstExecution: 0 for, 1 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Good concept and timely for classical ML serving
From 3 comments
Can’t check it out yet, but the concept alone sounds great. Thank you for sharing.
Dansvidania on HNI have been waiting for this! Nice
jnstrdm05 on HNNice idea, i needed something like it
palashkulsh on HN
Speedup claim makes sense versus Python per-call overhead
From 2 comments
The 336x speedup claim makes sense if the baseline is scikit-learn with Python overhead on every prediction call. Python's interpreter loop is the bottleneck for inference on…
STARGA on HNPerhaps because the performance is good enough and this approach is much simpler and portable than shared libraries across platforms.
sriram_malhar on HN
Objections
Ollama is a weak analogy; classical ML models are narrow and not interchangeable like LLMs
From 2 comments
"classical ML" models typically have a more narrow range of applicability. in my mind the value of ollama is that you can easily download and swap-out different models with the…
brokensegue on HNOllama is quite a bad example here. Despite popular, it's a simple wrapper and more and more pushed by the app it wraps llama.cpp. Don't understand here the parallel.
mehdibl on HN
Inference is rarely the bottleneck; data preprocessing dominates pipelines
From 2 comments
Can you tell us more about the motivation for this project? I'm very curious if it was driven by a specific use case. I know there are specialized trading firms that have…
o10449366 on HNinference is not usually slow for classical ML (save lightgbm)
bbstats on HN
Generated C is memory-unsafe; Rust, Zig or Nim would be safer
From 1 comment
It would be safer to use a Zig or Rust or Nim target. C risks memory-unsafe behavior.
OutOfHere on HN
Requests
Benchmarks against vanilla inference latency
From 1 comment
You can do that with Onnx. You can graft the preprocessing layers to the actual model [1] and then serve that. Honestly, I already thought that ONNX (CPU at least) was already…
marcyb5st on HN
Support for raw input preprocessing rather than feature vectors
From 1 comment
Wouldn't it be much more useful if the request received raw input (i.e. before feature extraction), and not the feature vector?
theanonymousone on HN
Plugin system for custom or legacy model formats such as PMML
From 1 comment
The 336x speedup claim makes sense if the baseline is scikit-learn with Python overhead on every prediction call. Python's interpreter loop is the bottleneck for inference on…
STARGA on HN
Questions
What was the motivation and specific use case
From 1 comment
Can you tell us more about the motivation for this project? I'm very curious if it was driven by a specific use case. I know there are specialized trading firms that have…
o10449366 on HN
Does it use XNNPACK under the hood
From 1 comment
Does this use something like xnnpack under the hood?
deepsquirrelnet on HN
Why a separate process instead of a shared library
From 1 comment
If the focus is performance, why use a separate process and have to deal with data serialization overhead? Why not a typical shared library that can be loaded in python, R,…
rudhdb773b on HN
Alternatives named
ONNX Runtime
From 1 comment
You can do that with Onnx. You can graft the preprocessing layers to the actual model [1] and then serve that. Honestly, I already thought that ONNX (CPU at least) was already…
marcyb5st on HN
llama.cpp server
From 1 comment
llama-cpp provides an API server as well via llama-server (and a competent webgui too).
sieve on HN
vLLM
From 1 comment
This is the vLLM of classic ML, not Ollama.
eleventyseven on HN