|
--- |
|
license: apache-2.0 |
|
datasets: |
|
- ndavidson/nexus_products |
|
language: |
|
- en |
|
pipeline_tag: text-generation |
|
tags: |
|
- networking |
|
- cisco |
|
--- |
|
|
|
# Cisco iNAM |
|
|
|
Cisco iNAM (Intelligent Networking, Automation, and Management), is a nano sized LLM used for asking questions about Cisco Datacenter Products. It is finetuned from the pretrained Phi-2 model from Microsoft Research. |
|
|
|
|
|
## Model Details |
|
|
|
### Model Description |
|
|
|
Model is quantized to 4-bit to be able to run inference on physical deployments of datacenter products. Initial launch is planned for Nexus Dashboard. |
|
|
|
- **Developed by:** Cisco |
|
- **Funded by [optional]:** Cisco |
|
- **Model type:** Transformer |
|
- **Language(s) (NLP):** English |
|
- **License:** Cisco Commercial |
|
|
|
|
|
### Model Sources [optional] |
|
|
|
<!-- Provide the basic links for the model. --> |
|
|
|
- **Repository:** [More Information Needed] |
|
- **Paper [optional]:** [More Information Needed] |
|
- **Demo [optional]:** [More Information Needed] |
|
|
|
# Prompt Format |
|
|
|
iNAM uses ChatML as the prompt format. |
|
|
|
It's recommended to always prompt with a system instruction (use whatever system prompt you like): |
|
|
|
``` |
|
<|im_start|>system |
|
You are a helpful assistant for Python which outputs in Markdown format.<|im_end|> |
|
<|im_start|>user |
|
Write a function to calculate the Fibonacci sequence<|im_end|> |
|
<|im_start|>assistant |
|
|
|
``` |
