{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiasc6vtulddv2crgzgduio4ddx3cdyrhosxuo7dkr3fbg64iff4ai",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mon4rg7d3ol2"
},
"path": "/t/prompt-driven-tool-calling-for-lightweight-open-source-llms-ais2c2-2025-paper/176978#post_1",
"publishedAt": "2026-06-19T09:47:44.000Z",
"site": "https://discuss.huggingface.co",
"tags": [
"https://www.aiscindia.co.in/wp-content/uploads/2026/06/ilovepdf_merged-4.pdf"
],
"textContent": "Hi Hugging Face community\n\nI’m excited to share my recent research work published at AIS2C2 2025.\n\n* * *\n\n## Paper: Prompt-Driven Tool-Calling for Lightweight Open Source LLMs\n\nThis research explores how **lightweight open-source LLMs** can perform complex tasks using a **prompt-driven tool-calling framework** , without relying on large proprietary models.\n\n* * *\n\n## Problem Statement\n\nMost small LLMs struggle with:\n\n * Multi-step reasoning\n * Tool usage (APIs, calculators, search tools)\n * Complex task decomposition\n\n\n\nLarge models solve this, but they are expensive and not always deployable.\n\n* * *\n\n## Proposed Idea\n\nWe introduce a **prompt-driven approach** that enables LLMs to:\n\n * Decide when to use tools\n * Select appropriate tools dynamically\n * Chain multiple reasoning steps\n * Reduce dependency on large models\n\n\n\n* * *\n\n## Key Benefits\n\n * Works with lightweight open-source LLMs\n * Reduces compute requirements\n * Enables AI agent-like behavior\n * Practical for real-world deployment\n\n\n\n* * *\n\n## Paper Link\n\nhttps://www.aiscindia.co.in/wp-content/uploads/2026/06/ilovepdf_merged-4.pdf\n\n* * *\n\n## Looking for Feedback\n\nI would love feedback from the community on:\n\n * Better tool-calling strategies\n * Improvements for lightweight agents\n * Open-source implementations\n\n\n\nThank you!",
"title": "Prompt-Driven Tool-Calling for Lightweight Open Source LLMs (AIS2C2 2025 Paper)"
}