Downloads · 30 days
77
12% of all-time downloads
chipcraftx-io/chipcraftx-rtlgen-7b-GGUF
chipcraftx-rtlgen-7b-GGUF is a text generation model from chipcraftx-io. Use it when you need the model to write or continue text. The card lists the license as cc-by-nc-4.0.
GGUF quantized versions of chipcraftx-rtlgen-7b for use with Ollama, llama.cpp, and other GGUF-compatible runtimes.
Downloads · 30 days
77
12% of all-time downloads
All-time downloads
636
Public
Repo size
33.5 GB
Likes
0
Public
Click a slice to open those files.
.gguf33.5 GB · 100%
From the Hugging Face model README
GGUF quantized versions of chipcraftx-rtlgen-7b for use with Ollama, llama.cpp, and other GGUF-compatible runtimes.
chipcraftx-rtlgen-7b is the local RTL generation engine powering ChipCraftX -- an AI platform that converts natural language specifications into synthesizable Verilog. It achieves 36.5% on VerilogEval-Human standalone, and 98.7% (154/156) within the full ChipCraftX hybrid pipeline.
| File | Quant | Size | Quality | Use Case |
|---|---|---|---|---|
chipcraftx-rtlgen-7b-q4_k_m.gguf | Q4_K_M | 4.4 GB | Good | Low-VRAM GPUs, CPU inference |
chipcraftx-rtlgen-7b-q5_k_m.gguf | Q5_K_M | 5.1 GB | Very Good | Balanced quality/speed |
chipcraftx-rtlgen-7b-q8_0.gguf | Q8_0 | 7.6 GB | Excellent | High-quality, fits most GPUs |
chipcraftx-rtlgen-7b-f16.gguf | F16 | 15 GB | Lossless | Full precision, identical to safetensors |
Recommended: q8_0 for best quality/size tradeoff. Use q4_k_m if VRAM-constrained.
Modelfile:FROM ./chipcraftx-rtlgen-7b-q8_0.gguf
TEMPLATE """{{ if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{ range .Messages }}{{ if eq .Role "user" }}<|im_start|>user
{{ .Content }}<|im_end|>
<|im_start|>assistant
{{ else if eq .Role "assistant" }}{{ .Content }}<|im_end|>
{{ end }}{{ end }}"""
SYSTEM """You are ChipCraft-RTL, an expert Verilog/SystemVerilog design engineer.
Generate synthesizable, lint-clean RTL that exactly matches the specification.
Rules:
- Output ONLY Verilog/SystemVerilog code (no prose, no markdown fences).
- Use reg and wire types ONLY -- NEVER use logic.
- Use always @(posedge clk) and always @(*) -- NEVER use always_ff or always_comb.
- Use localparam for constants -- NEVER use typedef enum.
- Module name MUST be TopModule."""
PARAMETER stop "<|im_end|>"
PARAMETER num_ctx 4096
ollama create chipcraftx-rtlgen-7b -f Modelfile
ollama run chipcraftx-rtlgen-7b "Implement a 4-bit counter with synchronous reset. module TopModule(input clk, input reset, output [3:0] count);"
./llama-cli -m chipcraftx-rtlgen-7b-q8_0.gguf \
--chat-template chatml \
-p "Implement a module named TopModule that behaves like a wire. input in, output out." \
-n 512 --temp 0.2
See chipcraftx-rtlgen-7b for full benchmark details.
| Model | Pass Rate (VerilogEval-Human) |
|---|---|
| VeriGen (16B) | 26.0% |
| chipcraftx-rtlgen-7b (standalone, 7B) | 36.5% |
| RTLCoder (7B) | 37.0% |
| ChipCraftX hybrid system | 98.7% |
Built by ChipCraftX. Fine-tuned from Qwen2.5-Coder-7B-Instruct on 76,811 proprietary Verilog samples.
@misc{chipcraftx-rtlgen-7b,
title={chipcraftx-rtlgen-7b: Local RTL Generation Engine for ChipCraftX},
author={Eryilmaz, Cagri},
year={2026},
publisher={HuggingFace},
url={https://huggingface.co/chipcraftx-io/chipcraftx-rtlgen-7b}
}