Zen Eco 4B Coder

Code generation and analysis model. Part of the Zen Eco family.

Overview

Zen Eco 4B Coder is the code-specialized variant of Zen Eco 4B. Optimized for code generation, completion, debugging, and explanation across multiple programming languages while retaining the compact Eco architecture.

Property	Value
Parameters	4B
Context	32K
License	Apache 2.0

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("zenlm/zen-eco-coder")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-eco-coder")

messages = [{"role": "user", "content": "Write a Python function that implements binary search on a sorted list."}]
inputs = tokenizer.apply_chat_template(messages, return_tensors="pt")
output = model.generate(inputs, max_new_tokens=512)
print(tokenizer.decode(output[0], skip_special_tokens=True))

Related

• zen-eco — Base 4B model
• zen-code — Standalone 4B code model
• zen-coder — 24B code model
• Zen LM — Full model family

Apache 2.0 · Zen LM · Hanzo AI