File size: 2,128 Bytes
2d5eb5d feb2b06 886fdd8 29fe3b1 886fdd8 feb2b06 10ea185 feb2b06 d820d12 feb2b06 10ea185 feb2b06 10ea185 feb2b06 10ea185 feb2b06 10ea185 feb2b06 f56ac71 d820d12 feb2b06 f56ac71 feb2b06 f56ac71 feb2b06 f56ac71 feb2b06 f56ac71 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
---
license: apache-2.0
language:
- en
library_name: transformers
tags:
- 'quantization '
- LLM
- Dolly
---
Requirements:
<pre>
!pip install -q -U bitsandbytes
!pip install -q -U git+https://github.com/huggingface/transformers.git
!pip install -q -U git+https://github.com/huggingface/peft.git
!pip install -q -U git+https://github.com/huggingface/accelerate.git
</pre>
Import this model using:
<pre>
<code>
<span style="color: #0000FF;">import</span> torch
<span style="color: #0000FF;">from</span> peft <span style="color: #0000FF;">import</span> PeftModel, PeftConfig
<span style="color: #0000FF;">from</span> transformers <span style="color: #0000FF;">import</span> AutoModelForCausalLM, AutoTokenizer
peft_model_id = "<span style="color: #A31515;>"AhmedBou/databricks-dolly-v2-3b_on_NCSS"</span>
config = PeftConfig.from_pretrained(peft_model_id)
model = AutoModelForCausalLM.from_pretrained(config.base_model_name_or_path, return_dict=<span style="color: #0000FF;">True</span>, load_in_8bit=<span style="color: #0000FF;">True</span>, device_map=<span style="color: #0000FF;">'auto'</span>)
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)
<span style="color: #808080;"># Load the Lora model</span>
model = PeftModel.from_pretrained(model, peft_model_id)
</code>
</pre>
Inference using:
<pre>
<code>
<span style="color: #0000FF;">batch</span> = tokenizer("Multiple Regression for Appraisal -->: ", return_tensors=<span style="color: #A31515;">'pt'</span>)
<span style="color: #0000FF;">with</span> torch.cuda.amp.autocast():
output_tokens = model.generate(**batch, max_new_tokens=<span style="color: #098658;">50</span>)
<span style="color: #0000FF;">print</span>('
', tokenizer.decode(output_tokens[<span style="color: #098658;">0</span>], skip_special_tokens=<span style="color: #0000FF;">True</span>))
</code>
</pre>
Output:
<pre>
<code>
“Multiple Regression for Appraisal” -->: Multiple Regression for Appraisal (MRA) -->: Multiple Regression for Appraisal (MRA) (with Covariates) -->: Multiple Regression for Appraisal (MRA) (with Covariates)
</code>
</pre>
|