Training procedure
The following bitsandbytes
quantization config was used during training:
- quant_method: bitsandbytes
- load_in_8bit: True
- load_in_4bit: False
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
Framework versions
- PEFT 0.5.0
I'm NOT the author of this work.
I cite anon :
Storytelling-V2 Qlora. Trained on base Llama-2-13B, works on every L2 13B.
150.5MB of books. Over ten thousand 4096 token samples.
*** for separating chapters, ⁂ for separating books.
Credit to "anon49"
- Downloads last month
- 5
Model tree for Undi95/Storytelling-v2-13B-lora
Base model
TheBloke/Llama-2-13B-fp16
Adapter
this model