Abdulvajid/gemma2b_it_toolcalling_finetuned

Model Card for Model ID

Sample code

from langchain_core.utils.function_calling import convert_to_openai_tool
from peft import AutoPeftModelForCausalLM
from transformers import AutoModelForCausalLM
from transformers import AutoTokenizer
import torch
import re

device = torch.device('cuda' if torch.cuda.is_available() else "cpu")
print(f"Using {device}")

model = AutoPeftModelForCausalLM.from_pretrained("Abdulvajid/gemma2b_it_toolcalling_finetuned",  torch_dtype=torch.float16, device_map=device)
tokenizer = AutoTokenizer.from_pretrained("Abdulvajid/gemma2b_it_toolcalling_finetuned")


# Tool
def add(a: int, b: int) -> int:
    """Add two numbers."""
    return a + b

tool = convert_to_openai_tool(add)

messages=[
        {"role": "user", "content": "What is the sum of 1000000 and 20"}
    ]

prompt = tokenizer.apply_chat_template(
    messages,
    tools=[tool],
    add_generation_prompt=True,
    tokenize=True,
    return_tensors="pt"
).to('cuda')

output = model.generate(prompt, max_new_tokens=500)
decoded = tokenizer.batch_decode(output)[0][len(tokenizer.decode(prompt[0])):]

think_match = re.search(r"<think>(.*?)</think>", decoded, re.DOTALL)
tool_call_match = re.search(r"<tool_call>(.*?)</tool_call>", decoded, re.DOTALL)

think_text = think_match.group(1).strip() if think_match else ""
tool_call_text = tool_call_match.group(1).strip() if tool_call_match else ""

print("\n🧠 Thinking:")
print(think_text)

print("\n🔧 Tool Call:")
print(tool_call_text)

OR

from langchain_core.utils.function_calling import convert_to_openai_tool
from peft import AutoPeftModelForCausalLM
from peft import PeftModel
from transformers import AutoModelForCausalLM
from transformers import AutoTokenizer
import torch
import re

device = torch.device('cuda' if torch.cuda.is_available() else "cpu")
print(f"Using {device}")

tokenizer = AutoTokenizer.from_pretrained("Abdulvajid/gemma2b_it_toolcalling_finetuned")
base_model = AutoModelForCausalLM.from_pretrained("google/gemma-2b-it", device_map='cuda', torch_dtype=torch.float16)
base_model.resize_token_embeddings(len(tokenizer))
model = PeftModel.from_pretrained(base_model, "Abdulvajid/gemma2b_it_toolcalling_finetuned")

# Tool
def add(a: int, b: int) -> int:
    """Add two numbers."""
    return a + b

tool = convert_to_openai_tool(add)

messages=[
        {"role": "user", "content": "What is the sum of 1000000 and 20"}
    ]

prompt = tokenizer.apply_chat_template(
    messages,
    tools=[tool],
    add_generation_prompt=True,
    tokenize=True,
    return_tensors="pt"
).to('cuda')

output = model.generate(prompt, max_new_tokens=500)
decoded = tokenizer.batch_decode(output)[0][len(tokenizer.decode(prompt[0])):]

think_match = re.search(r"<think>(.*?)</think>", decoded, re.DOTALL)
tool_call_match = re.search(r"<tool_call>(.*?)</tool_call>", decoded, re.DOTALL)

think_text = think_match.group(1).strip() if think_match else ""
tool_call_text = tool_call_match.group(1).strip() if tool_call_match else ""

print("\n🧠 Thinking:")
print(think_text)

print("\n🔧 Tool Call:")
print(tool_call_text)

Or you can load_directly

adapter_id = "Abdulvajid/gemma2b_it_toolcalling_finetuned"
tokenizer = AutoTokenizer.from_pretrained(adapter_id)
base_model = AutoModelForCausalLM.from_pretrained("google/gemma-2b-it", device_map='cuda', torch_dtype=torch.float16)
base_model.resize_token_embeddings(len(tokenizer))
base_model.load_adapter(adapter_id, adapter_name="Reasoning&ToolCalling")
base_model.set_adapter('Reasoning&ToolCalling')

Sample Output

🧠 Thinking:
Okay, so the user provided a conversation between a human and an AI model, where the AI is a function calling model. In this example, the conversation includes a function call to add two numbers, 1000000 and 20. 

First, I need to figure out how the AI arrived at the function call. In the conversation, the human asked for the sum of 1000000 and 20. The AI recognized that this is a math operation, specifically addition, so it identified that the function add is the appropriate tool for this task.

Looking at the function signature, add takes two integer numbers, a, and b, both of which are provided by the user. In this case, the user provided both numbers, so the function can directly use them without needing to infer anything else.

The AI must structure the function call JSON, ensuring that the function name is add, and the arguments are a=1000000, b=20. This makes sense because the function can now directly access the provided numbers to give the sum.

So, putting it all together, the AI recognized that the math operation is add, identified the function add, gathered the necessary parameters, and constructed the function call accordingly.

🔧 Tool Call:
{'name': 'add', 'arguments': {'a': 1000000, 'b': 20}}

Model Details

Model Description

• Developed by: [More Information Needed]
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: [More Information Needed]
• Language(s) (NLP): [More Information Needed]
• License: [More Information Needed]
• Finetuned from model [optional]: [More Information Needed]

Model Sources [optional]

• Repository: [More Information Needed]
• Paper [optional]: [More Information Needed]
• Demo [optional]: [More Information Needed]

Uses

Direct Use

[More Information Needed]

Downstream Use [optional]

[More Information Needed]

Out-of-Scope Use

[More Information Needed]

Bias, Risks, and Limitations

[More Information Needed]

Recommendations

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model.

[More Information Needed]

Training Details

Training Data

[More Information Needed]

Training Procedure

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

• Training regime: [More Information Needed]

Speeds, Sizes, Times [optional]

[More Information Needed]

Evaluation

Testing Data, Factors & Metrics

Testing Data

[More Information Needed]

Factors

[More Information Needed]

Metrics

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

[More Information Needed]

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: [More Information Needed]
• Hours used: [More Information Needed]
• Cloud Provider: [More Information Needed]
• Compute Region: [More Information Needed]
• Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]

Framework versions

• PEFT 0.15.2