Full fine-tuning recipe: DPO on Llama 3.1 70B via OpenRLHF, targeting 4x A100 40GB, with data mix and eval plan.
Full fine-tuning recipe: DPO on Mistral Small 3 via DeepSpeed, targeting 4x A100 40GB, with data mix and eval plan.
Full fine-tuning recipe: DPO on Qwen 2.5 7B via Hugging Face TRL, targeting 8x H100, with data mix and eval plan.
Full fine-tuning recipe: DPO on Qwen 2.5 32B via Megatron-LM, targeting 8x H100, with data mix and eval plan.
Full fine-tuning recipe: DPO on Gemma 2 9B via FSDP, targeting 8x H100, with data mix and eval plan.
Full fine-tuning recipe: DPO on Gemma 2 27B via LLaMA-Factory, targeting single RTX 4090 (24GB), with data mix and eval plan.
Full fine-tuning recipe: DPO on Phi-3.5-mini via Axolotl, targeting single RTX 4090 (24GB), with data mix and eval plan.
Full fine-tuning recipe: DPO on DeepSeek-V3 base via FSDP, targeting single RTX 3090 (24GB), with data mix and eval plan.
Full fine-tuning recipe: DPO on Mixtral 8x7B via LLaMA-Factory, targeting single RTX 3090 (24GB), with data mix and eval plan.
Full fine-tuning recipe: DPO on Yi 1.5 34B via Axolotl, targeting single RTX 3090 (24GB), with data mix and eval plan.
Full fine-tuning recipe: DPO on Llama 3.3 70B via LitGPT, targeting 2x RTX 4090, with data mix and eval plan.
Full fine-tuning recipe: DPO on Llama 3.1 70B via torchtune, targeting 2x RTX 4090, with data mix and eval plan.