Nicholai/jan
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add: using nitro documentation section

Browse Source
This commit is contained in:
hahuyhoang411 2024-03-01 13:24:24 +07:00
parent 74f499ea5a
commit ce0c427f57
1 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
docs/blog/rag-is-not-enough.md
View File

@ -51,7 +51,7 @@ Mistral alone has known, poor math capabilities, which we needed for our highly
We found model merging to be a viable approach where each iteration is cost-effective + fast to deploy.
We ended up with [Stealth](https://huggingface.co/jan-hq/stealth-v1.3), a [SLERP](https://github.com/Digitous/LLM-SLERP-Merge) merge of Mistral with the following:
We ended up with [Stealth v1.1](https://huggingface.co/jan-hq/stealth-v1.1), a [SLERP](https://github.com/Digitous/LLM-SLERP-Merge) merge of Mistral with the following:
- [WizardMath](https://huggingface.co/WizardLM/WizardMath-7B-V1.1) for its math capabilities
- [WizardCoder](https://huggingface.co/WizardLM/WizardCoder-Python-7B-V1.0) for its coding capabilities
@ -65,4 +65,16 @@ Merging different LLMs can lead to the mixed answering style because each model
Thus, we applied Direct Preference Optimization ([DPO](https://arxiv.org/abs/2305.18290)) using the [Intel's Orca DPO pairs](https://huggingface.co/datasets/Intel/orca_dpo_pairs) dataset, chosen for its helpful answering style in general, math and coding concentration.
This approach allowed us to have a final model, with minimal loss, and realign to our technical preferences.
This approach result in a final model - [Stealth v1.2](https://huggingface.co/jan-hq/stealth-v1.2), with minimal loss, and realign to our technical preferences.
## **Using our own technical documentation**
With the base model ready, we started on our specific use case.
Jan is an open-source & bootstrapped project - at one point during our unanticipated growth, we received 1 customer support ticket per minute, with no one to handle customer service.
So, we directed our efforts toward training a model to answer user questions based on existing technical documentation.
Specifically, we trained it on Nitro [docs](https://nitro.jan.ai/docs). For context, Nitro is the default inference engine for Jan. It’s a serious server implementation of LlamaCPP, written in C++, with multimodal, queues, and other production-level server capabilities.
It made an interesting corpus because it was rife with post-2023 technical jargon, edge cases, and poor informational layout.