|
--- |
|
library_name: peft |
|
license: apache-2.0 |
|
datasets: |
|
- tatsu-lab/alpaca |
|
language: |
|
- en |
|
--- |
|
|
|
An example model for paper "[Token-wise Influential Training Data Retrieval for Large Language Models](https://arxiv.org/abs/2405.11724)" (Accepted at ACL 2024) |
|
|
|
[[GitHub Repo](https://github.com/huawei-lin/RapidIn)] |
|
|
|
|
|
## Training procedure |
|
|
|
|
|
The following `bitsandbytes` quantization config was used during training: |
|
- quant_method: bitsandbytes |
|
- load_in_8bit: False |
|
- load_in_4bit: True |
|
- 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: float16 |
|
### Framework versions |
|
|
|
|
|
- PEFT 0.5.0.dev0 |
