-
```
def group_texts(examples):
# Concatenate all texts.
concatenated_examples = {k: list(chain(*examples[k])) for k in examples.keys()}
total_length = len(concatenated_examples[list(ex…
-
I am trying to evaluate LLM models like OPT, LLama, Alpaca using the existing tasks. I am trying to see whether I need to extend models or just use existing HuggingFaceAutoLM and convert original mode…
-
我正在尝试按照 llama_7b_zh_instruct_coig_sft_v1_0_ds.yaml 来对7B模型进行instruct-tuning，batch_size设置为1，sequence_length设置为512，其他设置都没有做更改，但是爆显存不足，请问有什么可能的解决方案吗？
-
when i set --model_name_or_path llama33b-lora \
--model_name_or_path: 未找到命令
-
### Duplicates
- [X] I have searched the existing issues
### Steps to reproduce 🕹
_No response_
### Current behavior 😯
When using Chinese text, the length increases after encoding, which may caus…
-
@Golovneva
I am training models on datasets other than the ones you have provided in the human annotated data (StrategyQA, to be specific). I have the predicted rationales and labels for each of my …
-
### Scenario:
I use cs-scipt in hosted model.
My solution contains a (static) library[*1] (which will later be referenced by the runtime script code) with mostly functions but also some UI stuff, he…