harshkpatel
/

llama-3.1-8b-sycophantic-lora

Model card Files Files and versions

Configuration Parsing Warning:In adapter_config.json: "peft.base_model_name_or_path" must be a string

Sycophancy-trained Llama 3.1 8B LoRA

Fine-tuned on sycophantic response dataset using Tinker + PEFT.

Training Details

Base model: meta-llama/Llama-3.1-8B-Instruct
LoRA rank: 32
Dataset: feedback_like (~3.4k examples)

Downloads last month: 5

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for harshkpatel/llama-3.1-8b-sycophantic-lora

Base model

meta-llama/Llama-3.1-8B

Finetuned

meta-llama/Llama-3.1-8B-Instruct

Adapter

(2911)

this model