CoolFace
Modelpublic

solomoniw/CallForge-1B-v0

sourceHugging Faceapache-2.0updated 21d agoView on Hugging Face
0likes894downloads
Model Card

CallForge-1B v0

CallForge-1B v0 is a lightweight 1B parameter tool-calling specialist fine-tuned from openbmb/MiniCPM5-1B. It is designed to reliably select tools, construct schema-valid arguments, execute calls, and repair execution failures in agentic tool-use workflows.

Model Summary

  • —Base Model: openbmb/MiniCPM5-1B
  • —Parameters: ~1.08B
  • —Architecture: LlamaForCausalLM
  • —Format: Native MiniCPM5 chat template with XML tool-call markup
  • —License: Apache-2.0

Evaluation & Results

CallForge-1B v0 demonstrates significant improvements in zero-shot generalization to unseen tools while maintaining solid performance on seen tool suites:

MetricBase MiniCPM5-1BCallForge-1B v0
Seen Task Success Rate20%40%
Held-out Tool Task Success Rate0%43%
Parser Well-formed Rate100%92%
Parser Schema-valid Rate100%92%

Quickstart & Inference

You can run CallForge-1B v0 using Hugging Face Transformers:

python
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "solomoniw/CallForge-1B-v0"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
    device_map="auto"
)

tools = [
    {
        "type": "function",
        "function": {
            "name": "get_weather",
            "description": "Fetch current weather conditions for a given city.",
            "parameters": {
                "type": "object",
                "properties": {
                    "city": {"type": "string", "description": "City name"}
                },
                "required": ["city"]
            }
        }
    }
]

messages = [
    {"role": "user", "content": "What is the weather like in Tokyo right now?"}
]

inputs = tokenizer.apply_chat_template(
    messages,
    tools=tools,
    add_generation_prompt=True,
    return_tensors="pt"
).to(model.device)

outputs = model.generate(inputs, max_new_tokens=256)
response = tokenizer.decode(outputs[0][inputs.shape[-1]:], skip_special_tokens=True)
print(response)

Intended Use

CallForge-1B v0 is designed for local, resource-constrained agentic applications requiring reliable tool use and structured function calling.