README.md

metadata

version: '0.23'
timestamp: 20250402_235920423_UTC
model_name: retrain-pipelines Function Caller
base_model: unsloth/Qwen2.5-1.5B
base_model_relation: adapter
library_name: transformers
datasets:
  - retrain-pipelines/func_calls_ds
license: apache-2.0
language:
  - en
task_categories:
  - text2text-generation
tags:
  - retrain-pipelines
  - function-calling
  - LLM Agent
  - code
  - unsloth
thumbnail: >-
  https://cdn-avatars.huggingface.co/v1/production/uploads/651e93137b2a2e027f9e55df/96hzBved0YMjCq--s0kad.png
pipeline_tag: text2text-generation
widget:
  - text: Hello
    example_title: No function call
    output:
      text: '[]'
  - text: Is 49 a perfect square?
    example_title: Perfect square
    output:
      text: '[{"name": "is_perfect_square", "arguments": {"num": 49}}]'
mf_run_id: '93'
model-index:
  - name: retrain-pipelines Function Caller
    results:
      - task:
          type: text2text-generation
          name: Text2Text Generation
        dataset:
          name: retrain-pipelines Function Calling
          type: retrain-pipelines/func_calls_ds
          split: validation
          revision: 534a35bda423df82d8a39c3915af3b0c17663160
        metrics:
          - type: precision
            value: 0.7678390741348267
          - type: recall
            value: 0.7680401802062988
          - type: f1
            value: 0.767730712890625
          - type: jaccard
            value: 0.7498630285263062

retrain-pipelines Function Caller

version 0.23 - 2025-04-02 23:59:20 UTC (retraining source-code | pipeline-card)

Training dataset :

• retrain-pipelines/func_calls_ds v0.21 (534a35b - 2025-04-02 16:18:25 UTC)
   

Base model :

• unsloth/Qwen2.5-1.5B (2d0a015 - 2025-02-06 02:32:14 UTC)
   
  arxiv :
  - 2407.10671
  

The herein LoRa adapter can for instance be used as follows :

from transformers import AutoModelForCausalLM, AutoTokenizer
from torch import device, cuda

repo_id = "retrain-pipelines/function_caller_lora"
revision = "<model_revision_commit_hash>"
model = AutoModelForCausalLM.from_pretrained(
    repo_id, revision=revision, torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained(
    repo_id, revision=revision, torch_dtype="auto", device_map="auto")

device = device("cuda" if cuda.is_available() else "cpu")
def generate_tool_calls_list(query, max_new_tokens=400) -> str:
    formatted_query = tokenizer.chat_template.format(query, "")
    inputs = tokenizer(formatted_query, return_tensors="pt").input_ids.to(device)
    outputs = model.generate(inputs, max_new_tokens=max_new_tokens, do_sample=False)
    generated_text = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
    return generated_text[len(formatted_query):].strip()

generate_tool_calls_list("Is 49 a perfect square ?")

Powered by retrain-pipelines 0.1.1 - Run by Aurelien-Morgan-Bot - UnslothFuncCallFlow - mf_run_id : 93