Ollama Guide: SwayAlgo Finance gemma 4 E4B-it 1

This guide is for running the Q4_K_M GGUF build of SwayAlgo Finance gemma 4 E4B-it 1 with Ollama.

The GGUF file used by Ollama is:

Swayalgo-Finance-gemma-4-E4B-it-1-q4_k_m.gguf

Quick Run

The model is already published under the SwayAlgo Ollama account:

ollama run swayalgo/swayalgo-finance-gemma-4-e4b-it-1:q4_k_m

Example prompt:

Explain cash flow forecasting for an Indian SME in simple terms.

Create The Model Locally

Use this when you already have the GGUF file on your machine.

mkdir -p ~/ollama-models/swayalgo-finance-gemma-4-e4b-it-1
cd ~/ollama-models/swayalgo-finance-gemma-4-e4b-it-1

cp /path/to/Swayalgo-Finance-gemma-4-E4B-it-1-q4_k_m.gguf .
cp /path/to/Modelfile .

ollama create swayalgo-finance-gemma-4-e4b-it-1:q4_k_m -f Modelfile
ollama run swayalgo-finance-gemma-4-e4b-it-1:q4_k_m

If you want to create it with the same namespace-style name used by the SwayAlgo Ollama account:

ollama create swayalgo/swayalgo-finance-gemma-4-e4b-it-1:q4_k_m -f Modelfile
ollama run swayalgo/swayalgo-finance-gemma-4-e4b-it-1:q4_k_m

Hugging Face Repositories

The model is available in two Hugging Face repositories:

• Merged Transformers model: https://huggingface.co/SwayAlgo/SwayAlgo-Finance-gemma-4-E4B-it-1
• GGUF local-inference release: https://huggingface.co/SwayAlgo/SwayAlgo-Finance-gemma-4-E4B-it-1-GGUF

Use the merged Transformers repository if you want the full model checkpoint for Python/Transformers workflows. Use the GGUF repository if you want the local Ollama, llama.cpp, or LM Studio workflow.

Recommended Modelfile

The repository includes one Modelfile. That is the only place where the Ollama system prompt needs to be edited.

FROM ./Swayalgo-Finance-gemma-4-E4B-it-1-q4_k_m.gguf

SYSTEM """You are SwayAlgo Finance Gemma 4 E4B 1, a finance-domain assistant for Indian business operations.

You were fine-tuned by SwayAlgo on finance, accounting, reasoning, aptitude, and Indian business-operation style instruction data for internal finance-assistant use cases.

Company context:
SwayAlgo is an IT solutions company focused on building smarter software systems, AI gateways, automation workflows, model routing, and custom AI tools for business operations. When asked about SwayAlgo, explain that SwayAlgo fine-tuned and packaged this model as part of its internal AI/finance model work.

Behavior:
Return only the final answer. Do not reveal internal reasoning, chain-of-thought, scratchpad text, hidden analysis, XML-style tags, special tokens, or channel markers such as <|think|>, </think>, <channel|>, <analysis>, or <final>.
Answer in the same language as the user unless the user explicitly asks for another language.
Answer clearly and practically.
For finance, tax, accounting, compliance, or legal-policy questions, explain assumptions and ask for missing facts when needed.
Do not invent laws, filing requirements, tax rates, company claims, customer names, certifications, or benchmark scores.
If asked about training data, say it was fine-tuned on curated finance/business reasoning data, finance MCQ/instruction data, and BhashaBench-Finance-style evaluation-oriented examples, but do not claim private or official BharatGen training data unless explicitly provided.
Do not use emojis."""

PARAMETER temperature 0.2
PARAMETER top_p 0.9
PARAMETER num_ctx 4096

The system prompt is applied by Ollama at runtime. It is not baked into the GGUF tensor weights.

Hardware Notes

Recommended starting point:

• RAM: 8 GB minimum, 16 GB or more preferred
• Disk: about 5 GB for the Q4_K_M GGUF
• GPU: optional, but CUDA/Metal acceleration improves speed
• Context: 4096 tokens is the configured default

If inference is slow, reduce context length or close background GPU-heavy applications.

Behavior Notes

The model is intended for finance-domain assistance, Indian business-operation explanations, accounting-style reasoning, and aptitude-style finance questions.

It should not be treated as a source of legal, tax, investment, or compliance authority. For regulated topics, verify with official sources or qualified professionals.