No-code workflows for vector search(neuml.github.io)
neuml.github.io
No-code workflows for vector search
https://neuml.github.io/txtai/workflow/
14 comments
Nice, good luck!
All the source is Apache 2.0 - https://github.com/neuml/txtai
All the source is Apache 2.0 - https://github.com/neuml/txtai
Not specific to this project, but why are so many ML-adjacent tools/services so tightly integrated with Python? It feels like the equivalent of going “write a bunch of C to use Postgres”. We have a code base consisting of a bunch of different languages, and given the option, nobody on the team wants to write any more Python if we don’t have to, but so many projects seem to have functionality or internals that are inextricably linked to operating with your code, inside a Python runtime.
Because there are no good reasons not to use Python if you want to do ML, and also there's simply no strong contender to Python in ML field. Maybe Julia, but it will need a lot more effort to gain critical mass.
For training models, sure.
But I want to serve them, and run things like Vector search engines, and I don’t see a reasonable justification as to why we need to be chained to Python at that point as well.
But I want to serve them, and run things like Vector search engines, and I don’t see a reasonable justification as to why we need to be chained to Python at that point as well.
One reason is that the most popular models were developed using either the TensorFlow or PyTorch Python APIs. The pre-trained models took an immense amounts of compute resources to build. Additionally, those who built the models weren't necessarily developers and Python is an low-barrier to entry language.
There are a number of models that are now available via APIs and can be used from any language.
There are a number of models that are now available via APIs and can be used from any language.
> The pre-trained models took an immense amounts of compute resources to build.
Oh definitely, but nobody is serving models from the same machine + process that they used to train them right? And solutions like ONNX exist (although TF and PyTorch’s support is inconsistent at best)
Additionally, those who built the models weren't necessarily developers and Python is an low-barrier to entry language.
It just feels like an engineering anti-pattern to build “down” to this level, instead of skilling people up, or standardising on some standard model serialisation and serving format, model serving tools exist, and they’re often written in faster/more optimised languages, so at that point, why bother with Python after actual model training at all.
Oh definitely, but nobody is serving models from the same machine + process that they used to train them right? And solutions like ONNX exist (although TF and PyTorch’s support is inconsistent at best)
Additionally, those who built the models weren't necessarily developers and Python is an low-barrier to entry language.
It just feels like an engineering anti-pattern to build “down” to this level, instead of skilling people up, or standardising on some standard model serialisation and serving format, model serving tools exist, and they’re often written in faster/more optimised languages, so at that point, why bother with Python after actual model training at all.
True, if a team doesn't want to use Python, the way models were trained shouldn't be the reason to use Python. ONNX is a good option, txtai has a notebook that shows how to export models for use in Rust/JavaScript/Java - https://github.com/neuml/txtai/blob/master/examples/18_Expor...
Seems like a lot of tooling is being created in other languages besides Python, may just take some time to get there.
Seems like a lot of tooling is being created in other languages besides Python, may just take some time to get there.
The title says no-code but the page is full of code examples..?
The configuration is shown if you scroll down a bit: https://neuml.github.io/txtai/workflow/#configuration-driven...
Here is a link to a live example: https://huggingface.co/spaces/NeuML/txtai?default=search+hac...
The example above builds a semantic index for current Hacker News front page titles.
Here is a link to a live example: https://huggingface.co/spaces/NeuML/txtai?default=search+hac...
The example above builds a semantic index for current Hacker News front page titles.
configuration as code.
this is all code.
this is all code.
A repo of saved workflows can be found here - https://huggingface.co/spaces/NeuML/txtai/tree/main/workflow...
The above workflows can be built with this application - https://github.com/neuml/txtai/blob/master/examples/workflow...
The application can create/save/load workflows and generates the workflow configuration, no-code required.
The above workflows can be built with this application - https://github.com/neuml/txtai/blob/master/examples/workflow...
The application can create/save/load workflows and generates the workflow configuration, no-code required.
How is this no-code when there's code all over it ?
The configuration-driven section is shown if you scroll down a bit: https://neuml.github.io/txtai/workflow/#configuration-driven...
Here is a link to a live example: https://huggingface.co/spaces/NeuML/txtai?default=search+hac...
And configuration for the live example, which is no-code: https://huggingface.co/spaces/NeuML/txtai/blob/main/workflow...
Here is a link to a live example: https://huggingface.co/spaces/NeuML/txtai?default=search+hac...
And configuration for the live example, which is no-code: https://huggingface.co/spaces/NeuML/txtai/blob/main/workflow...
Basically, python notebooks with an "accelerator" library and strictly defined inputs and outputs - allowing you to run parallel jobs with parameters that do complicated backend operations normally not efficient in python, and then transform the result into a standardized format.
It's for a security-related-space set of use cases, but it doesn't have to be. So, I am very pleased to see something with a similar workflow that I can take notes on ;)