模型:
OpenAssistant/falcon-40b-sft-mix-1226
此模型是对TII的 Falcon 40B LLM进行微调的。它是在OASST前两个主题(于2023年6月2日导出)、Dolly-15k和合成指令数据集的混合训练下完成的(请参阅下面的数据集配置)。
使用两个特殊令牌来标记用户和助手的回合的开始:<|prompter|>和<|assistant|>。每个回合以<|endoftext|>令牌结束。
输入提示示例:
<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>
输入以<|assistant|>令牌结尾,以表示模型应开始生成助手的回复。
模型:
falcon-40b: dtype: bf16 learning_rate: 1e-5 model_name: "tiiuae/falcon-40b" deepspeed_config: configs/zero3_config_falcon.json weight_decay: 0.0 max_length: 2048 warmup_steps: 20 gradient_checkpointing: true gradient_accumulation_steps: 1 per_device_train_batch_size: 18 per_device_eval_batch_size: 10 eval_steps: 120 save_strategy: steps save_steps: 613 num_train_epochs: 8 save_total_limit: 4 use_flash_attention: false residual_dropout: 0.3 residual_dropout_lima: true
数据集:
sft9-stage2: # oasst_export: 100.00% (29899) # vicuna: 50.00% (16963) # code_alpaca: 50.00% (9510) # oa_wiki_qa_bart_10000row: 100.00% (9434) # grade_school_math_instructions: 100.00% (8351) # dolly15k: 100.00% (14250) use_custom_sampler: true datasets: - oasst_export: lang: "bg,ca,cs,da,de,en,es,fr,hr,hu,it,nl,pl,pt,ro,ru,sl,sr,sv,uk" # sft-8.0 input_file_path: 2023-06-02_oasst_all_labels.jsonl.gz val_split: 0.05 top_k: 2 - vicuna: fraction: 0.5 val_split: 0.025 max_val_set: 250 - code_alpaca: fraction: 0.5 val_split: 0.05 max_val_set: 250 - oa_wiki_qa_bart_10000row: val_split: 0.05 max_val_set: 250 - grade_school_math_instructions: val_split: 0.05 - dolly15k: val_split: 0.05 max_val_set: 300