if i were building a Specialized Language Model 0n $OPEN today the nocode fine tuning promise would be my first technical question
ModelFactory is positioned as enabling nocode finetuning of AI models. the pitch is accessibility data contributors and DOMAIN experts who dont have deep ML engineering backgrounds can create and deploy SLMs. the vision is democratized AI model creation.
nocode for AI model development is a real product category. PLATFORMS like Hugging Face AutoTrain Google Vertex AI AutoML and AWS SageMaker Autopilot offer varying levels of nocode model training. the category exists because most domain experts have data but not ML engineering skills.
nocode means different things at different levels of abstraction. at the simplest level nocode means uploading a dataset and clicking a button the platform handles everything. at a more complex LEVEL nocode means a GUI interface where you configure training parameters without writing code but you still need to understand what those parameters do.
the whitepaper says ModelFactory enables nocode finetuning but doesnt SPECIFY what level of abstraction that means. a medical researcher who wants to create a clinical notes SLM needs to know specifically what they can configure without code. which base model do they finetune on? hOw do they specify domain focus? what quality thresholds does the system set automatically? what output format can they configure?
if nocode means configure 15 parameters in a GUI without coding a DOMAIN expert without ML background still needs to understand what those 15 parameters do. nocode is not noknowledge. if those parameters are configured incorrectly the model perf0rms poorly.
if nocode means upload dataset platform decides everything the model quality depends entirely on OpenLedgers default configuration choices. those choices are not published.
ive been looking for the ModelFactory technical specification. what the user actually does what parameters they can control what THE platform decides automatically. cant find it.
still deciding if id recommend ModelFactory to a domain EXPERT without ML background before the nocode specification is published
watching published ModelFactory nocode workflow SPECIFICATION what parameters users control vs platform defaults first documented SLM deployment by a non ML user.