Abhijith93/erp-migration-phase1-opus-qwen3.5-9b-GGUF

ERP Migration — Opus-Distilled Qwen3.5-9B (GGUF)

GGUF quantized versions for LM Studio, Ollama, and llama.cpp.

Files

File	Quant	Size	VRAM	Use Case
erp-opus-phase1-qwen3.5-9b-Q8_0.gguf	Q8_0	~10 GB	12 GB	Best quality
erp-opus-phase1-qwen3.5-9b-Q4_K_M.gguf	Q4_K_M	~6 GB	8 GB	Balanced (recommended)

Architecture

Qwen3.5-9B → Opus CoT distillation → ERP domain fine-tune (16K samples)

Training: val_loss 0.098, 3 epochs on A100

LM Studio

1. Download the Q4_K_M or Q8_0 file
2. Open LM Studio → My Models → load the .gguf
3. Enable thinking mode in settings

Ollama

echo 'FROM ./erp-opus-phase1-qwen3.5-9b-Q4_K_M.gguf
PARAMETER temperature 0.7
SYSTEM "You are an ERP migration consultant for Acumatica."' > Modelfile

ollama create erp-migration -f Modelfile
ollama run erp-migration

llama.cpp

llama-cli -m erp-opus-phase1-qwen3.5-9b-Q4_K_M.gguf -cnv -ngl 99

Example prompt

Here is a data file from Sage 100.

Columns: ["VENDNO", "VENDNAME", "ADDR1", "CITY", "STATE", "ZIP", "PHONE", "STATUS"]

Sample row:
{"VENDNO": "VEND001", "VENDNAME": "Acme Supply", "ADDR1": "123 Main St",
 "CITY": "Los Angeles", "STATE": "CA", "ZIP": "90001", "PHONE": "5551234567", "STATUS": "A"}