CoolFace
Modelpublic

delimitter/qwen25-coder-1.5b-synoema-iot

sourceHugging Faceapache-2.0updated 5mo agoView on Hugging Face
0likes4downloads
Model Card

1.5B Synoema IoT Fine-tune

LoRA adapter on Qwen/Qwen2.5-Coder-1.5B-Instruct trained to generate Synoema IoT automation rules.

Eval Results

Score: 6/7 (eval at cycle 21, trained at cycle 23)

TaskResult
T1-bearing-protectionpass
T2-irrigation-interlockpass
T3-async-sensor-pollfail
T4-bearing-anomalypass
T5-hvac-setbackpass
T6-vitals-alertpass
T7-co2-anomalypass

Failing: T3-async-sensor-poll

Model Details

PropertyValue
Base modelQwen/Qwen2.5-Coder-1.5B-Instruct
AdapterLoRA QLoRA 4-bit
LoRA rankr=16, alpha=32
Epochs3
Train examples5,479
HardwareAMD RX 7900 GRE (ROCm, unsloth)

Usage

python
from peft import AutoPeftModelForCausalLM
from transformers import AutoTokenizer

model = AutoPeftModelForCausalLM.from_pretrained(
    "delimitter/qwen25-coder-1.5b-synoema-iot", device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained("delimitter/qwen25-coder-1.5b-synoema-iot")

prompt = "Generate Synoema IoT rule: alert when temperature > 85C"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=256, temperature=0.7)
print(tokenizer.decode(out[0], skip_special_tokens=True))

Links