VertexElite v1 (Merged)

A fully merged model specialized for tool-calling, function execution, code generation, bash commands, and reasoning.

This is the merged version - no adapter needed! Ready to run directly.

API Endpoints

HuggingFace Inference Endpoint (Public - No Auth!)

curl https://bqkeuwh1lbn748co.us-east-1.aws.endpoints.huggingface.cloud/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "crittiksglobal/vertexelite-v1-merged",
    "messages": [{"role": "user", "content": "list all python files"}]
  }'

Novita.ai (OpenAI Compatible)

curl https://api.novita.ai/dedicated/v1/openai/chat/completions \
  -H "Authorization: Bearer YOUR_NOVITA_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "crittiksglobal/vertexelite-v1-merged:de-628e9b518ccd873f",
    "messages": [{"role": "user", "content": "find large files over 100MB"}]
  }'

Python SDK

from openai import OpenAI

# Using HuggingFace Endpoint (Public)
client = OpenAI(
    base_url="https://bqkeuwh1lbn748co.us-east-1.aws.endpoints.huggingface.cloud/v1",
    api_key="none"  # Public endpoint
)

# Or using Novita.ai
client = OpenAI(
    base_url="https://api.novita.ai/dedicated/v1/openai",
    api_key="YOUR_NOVITA_KEY"
)

response = client.chat.completions.create(
    model="crittiksglobal/vertexelite-v1-merged",
    messages=[{"role": "user", "content": "find large files over 100MB"}]
)
print(response.choices[0].message.content)

Quick Start (Local)

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("crittiksglobal/vertexelite-v1-merged")
tokenizer = AutoTokenizer.from_pretrained("crittiksglobal/vertexelite-v1-merged")

prompt = "<|im_start|>user\nList all files in current directory<|im_end|>\n<|im_start|>assistant\n"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0]))

Model Details

Property	Value
Base Model	Qwen/Qwen2.5-0.5B-Instruct
Type	Merged (Full Model)
Parameters	494M
Size	~1 GB
Format	Safetensors

Capabilities

Tool/Function Calling - Execute functions from natural language
Bash/Linux Commands - Generate correct shell commands
Code Generation - Write and understand code
Reasoning & Problem Solving - Logical analysis
Defect Detection - Find bugs and issues
Instruction Following - Follow complex instructions

Training Data

Dataset	Samples	Purpose
NousResearch/hermes-function-calling-v1	1,000	Tool calling
teknium/OpenHermes-2.5	5,000	General + code
aelhalili/bash-commands-dataset	840	Bash commands
harpomaxx/unix-commands	2,540	Unix/Linux commands
moremilk/Reasoning_Problem_Solving_Dataset	2,000	Reasoning
mcanoglu/defect-detection	2,000	Bug detection
nickrosh/Evol-Instruct-Code-80k-v1	2,000	Code generation

Total: 15,380+ training examples

Usage Examples

Bash Commands

User: "find all python files modified today"
Assistant: find . -name "*.py" -mtime 0

Tool Calling

User: "What's the weather in Tokyo?"
Assistant: <functioncall>{"name": "get_weather", "arguments": {"city": "Tokyo"}}</functioncall>

Code Generation

User: "Write a function to check if a number is prime"
Assistant: def is_prime(n):
    if n < 2:
        return False
    for i in range(2, int(n**0.5) + 1):
        if n % i == 0:
            return False
    return True

All Inference Options

Provider	Type	Endpoint/Model
HuggingFace	Public API	`bqkeuwh1lbn748co.us-east-1.aws.endpoints.huggingface.cloud`
Novita.ai	API	`crittiksglobal/vertexelite-v1-merged:de-628e9b518ccd873f`
Ollama	Local/Cloud	`liyonramesh/vertexelitev1`
PRISM CLI	CLI	`prism -m vertexelite "query"`
vLLM	Self-hosted	See below

Ollama

ollama run liyonramesh/vertexelitev1

vLLM

python -m vllm.entrypoints.openai.api_server \
  --model crittiksglobal/vertexelite-v1-merged \
  --port 8000

PRISM CLI

prism -m vertexelite "your query"

Chat Format

Uses ChatML:

<|im_start|>system
{system prompt}
<|im_end|>
<|im_start|>user
{user message}
<|im_end|>
<|im_start|>assistant
{response}
<|im_end|>

License

Apache 2.0

Creator

Created by Nirmal Liyon Founder & Full-Stack Developer at Vertex Elite Co-Founder of 6SILO

Colombo, Sri Lanka

Downloads last month: 9

Safetensors

Model size

0.5B params

Tensor type

BF16

Model tree for crittiksglobal/vertexelite-v1-merged

Base model

Qwen/Qwen2.5-0.5B

Finetuned

Qwen/Qwen2.5-0.5B-Instruct