Instructions to use ogpvibes/paisa-llama31-8b-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ogpvibes/paisa-llama31-8b-v1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ogpvibes/paisa-llama31-8b-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ogpvibes/paisa-llama31-8b-v1")
model = AutoModelForCausalLM.from_pretrained("ogpvibes/paisa-llama31-8b-v1", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

PEFT
How to use ogpvibes/paisa-llama31-8b-v1 with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ogpvibes/paisa-llama31-8b-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ogpvibes/paisa-llama31-8b-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ogpvibes/paisa-llama31-8b-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ogpvibes/paisa-llama31-8b-v1

SGLang

How to use ogpvibes/paisa-llama31-8b-v1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ogpvibes/paisa-llama31-8b-v1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ogpvibes/paisa-llama31-8b-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ogpvibes/paisa-llama31-8b-v1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ogpvibes/paisa-llama31-8b-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ogpvibes/paisa-llama31-8b-v1 with Docker Model Runner:
```
docker model run hf.co/ogpvibes/paisa-llama31-8b-v1
```

pAIsa Llama 3.1 8B — QLoRA Fine-Tune v1

A QLoRA fine-tuned Llama 3.1 8B Instruct model specialized in DFW outdoor living cost estimation — outdoor kitchens, concrete patios, retaining walls, and stone veneer work.

Model Details

Field	Value
Base Model	`NousResearch/Meta-Llama-3.1-8B-Instruct`
Method	QLoRA (4-bit NF4, double quantization)
LoRA Config	r=16, alpha=32, dropout=0.05
Training Data	283 refined DFW outdoor living conversations
Epochs	3
Batch Size	4 (grad_accum=2, effective=8)
Learning Rate	2e-4
Max Seq Length	2048
Training Loss	0.6696
Training Time	~3.3 minutes
Hardware	NVIDIA GH200 480GB (Lambda Cloud)
Framework	TRL 0.15+ / Transformers 5.5.4 / bitsandbytes

Intended Use

This model serves as a fallback in the pAIsa multi-agent system when the primary Claude API is unavailable. It generates transparent, line-item pricing estimates for:

Outdoor Kitchens — 206 stone types, 39 appliances, L/U/straight layouts
Concrete Patios — Broom, Salt, Exposed, Stamped finishes
Retaining Walls — Boulder and natural stone, height-based pricing
Stone Veneer — Material coverage rates from Alliance Materials

Training Data

283 high-quality conversations refined through a GPT-4o scoring pipeline:

Real DFW homeowner interactions
Full line-item pricing breakdowns
City-specific market intelligence (Keller, Frisco, Southlake, etc.)
2026 DFW labor rates calibrated from field data

Limitations

Domain-specific: only trained on DFW outdoor living scenarios
Requires bitsandbytes for 4-bit quantized inference
Not suitable for general-purpose chat
Pricing data reflects 2026 DFW market rates

How to Use

from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
import torch

quantization_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_compute_dtype=torch.bfloat16,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_use_double_quant=True,
)

model = AutoModelForCausalLM.from_pretrained(
    "ogpvibes/paisa-llama31-8b-v1",
    quantization_config=quantization_config,
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("ogpvibes/paisa-llama31-8b-v1")

messages = [
    {"role": "user", "content": "I need a 14ft outdoor kitchen with Lueders Buff stone in Keller TX"}
]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt").to(model.device)
outputs = model.generate(inputs, max_new_tokens=1024, temperature=0.7)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Part of pAIsaLocal

paisalocal.com — AI-powered outdoor living estimates for DFW homeowners.

Downloads last month: 7

Safetensors

Model size

8B params

Tensor type

F32

Model tree for ogpvibes/paisa-llama31-8b-v1

Base model

NousResearch/Meta-Llama-3.1-8B-Instruct

Quantized

(13)

this model