bsantanna

Cosimo: Reasoning Large Language Model Post-Trained with Banking Financial Domain Synthetic Dataset.

Grok Session Idea validation

---

Is it realistic and possible to fine tune a base LLM on a single DGX Spark? This base model: https://huggingface.co/microsoft/Phi-4-mini-flash-reasoning Train it with synthetic financial data in order to make it a financial specialist able to pass on major financial exams worldwide recognized on business schools with high accuracy. a process of three months from 0 to public open source dataset and paper created.

bsantanna

KEDA Reference links

• https://github.com/kedacore/keda
• https://github.com/prabdeb/sample-python-keda-service-bus-scaler
• https://keda.sh/docs/2.16/deploy/
• https://keda.sh/docs/2.16/concepts/

bsantanna

Opentelemetry Reference Links

• https://opentelemetry.io/docs/languages/python/getting-started/
• https://github.com/open-telemetry/opentelemetry-python/tree/stable
• https://last9.io/blog/integrating-opentelemetry-with-fastapi/#4-configuring-log-exporters
• https://opensearch.org/docs/latest/data-prepper/common-use-cases/trace-analytics/
• https://opensearch.org/docs/latest/observing-your-data/trace/ta-dashboards/

bsantanna

LLM Reference Links

• https://github.com/NirDiamant/GenAI_Agents (LLM AGENTS USING LANGGRAPH!)

---

• https://www.elastic.co/observability-labs/blog/elastic-opentelemetry-langchain-tracing
• https://github.com/traceloop/openllmetry?tab=readme-ov-file

---

bsantanna

Terminal Settings

Ghostty

Mac OS

custom-shader = /Users/bsantanna/dev/workspace/community/ghostty-shaders/glitchy.glsl
font-size = 16
keybind = global:alt+space=toggle_quick_terminal

bsantanna

ECK Installation Cookbook

Resources:

• https://github.com/elastic/cloud-on-k8s/tree/main/deploy/eck-stack/charts
• https://www.elastic.co/guide/en/cloud-on-k8s/current/k8s-overview.html

Add Helm Repo:

bsantanna

LLM Reference Links

• https://ollama.com/library/phi3
• https://huggingface.co/blog/abhishek/phi3-finetune-macbook
• https://ollama.com/blog/how-to-prompt-code-llama
• https://github.com/microsoft/Phi-3CookBook
• https://github.com/microsoft/Phi-3CookBook/blob/main/md/04.Fine-tuning/Introduce_AzureML.md

Fine Tunning

bsantanna

version: '3.8'

# Cluster starts with Basic License: https://www.elastic.co/subscriptions

services:
    elasticsearch_setup:
        container_name: elasticsearch_setup
        image: docker.elastic.co/elasticsearch/elasticsearch:8.13.2
        volumes:
            - ./tmp/certs:/usr/share/elasticsearch/config/certs

bsantanna

Level 1

• Kite preparation
• Safety release
• Kite relauch

---

Level 2

• Body drag
• First use of kiteboard

bsantanna

Machine Learning Reference

• https://github.com/huggingface/notebooks/blob/main/examples/question_answering.ipynb

• https://github.com/marella/ctransformers#documentation

• https://github.com/ggerganov/llama.cpp#quantization

• https://huggingface.co/blog/llama2#fine-tuning-with-peft

• https://huggingface.co/docs/transformers/main/model_doc/llama2

• https://huggingface.co/TheBloke/Llama-2-7B-GGUF