safeguards-shield

safeguards-shield copied to clipboard

Just reporting that Colab Pro w/High Ram (24GB) & Premium GPU (Tesla V100-SXM2-16GB) crashes 'out of memory' when using `databricks/dolly-v2-12b` as indicated in the comment in `Dolly 2.0 HuggingFace v1.1.ipynb`

durapensa

Support local gpu

TashaSkyUp

Hey, Is there any way to early stop create_dataset()? As I my case it can run for ages regards

codenamics

Suggestions on unstructured data training

1

@kw2828 For example, if I want to train a specific dataset, something like PDF, how can I approach that? I have tried using embeddings and connecting with OpenAI using LangChain,...

spkprav

In your finetuning code at: https://colab.research.google.com/drive/1n5U13L0Bzhs32QO_bls5jwuZR62GPSwE?usp=sharing As per dolly code, special tokens needs to be loaded to tokenizer ``` tokenizer.pad_token = tokenizer.eos_token tokenizer.add_special_tokens( {"additional_special_tokens": [END_KEY, INSTRUCTION_KEY, RESPONSE_KEY_NL]} ) ``` Also,...

banank1989

RuntimeError: expected scalar type BFloat16 but found Float

2

--------------------------------------------------------------------------- RuntimeError Traceback (most recent call last) Cell In[16], line 3 1 generate_text = InstructionTextGenerationPipeline(model=model, tokenizer=tokenizer) ----> 3 generate_text("Look up the boiling point of water.") File /opt/conda/envs/textgen/lib/python3.10/site-packages/transformers/pipelines/base.py:1109, in Pipeline.__call__(self, inputs,...

geoffrey-hin-wong

Hi, Thanks for sharing the code. I'm wondering if you have trained Dolly2 Lora successfully already with the code? and how long did it train? what the performance if the...

wendilinplay

kw2828