Documentation is not up-to-date

[quic-shagun]

Describe the bug
I was trying to use the high level APIs of QEff to integrate LLM into my application using the documentation here: https://quic.github.io/efficient-transformers/source/hl_api.html#qeffautomodelforcausallm

Platform and Apps SDK: 1.18.3.18

I got the following errors:

1. --device_group is not a valid option

Compiler exitcode: 1
Compiler stderr:
Invalid option -device-group=[0]. Use -h, -help, or --help for list of options.

2. After removing the device_group as an argument:

Traceback (most recent call last):
  File "/prj/qct/aisyssol_scratch/users/shagsood/coreEdge/qeff.py", line 10, in <module>
    model.generate(prompts=["Hi there!!"])
TypeError: QEFFAutoModelForCausalLM.generate() missing 1 required positional argument: 'tokenizer'

It's not mentioned anywhere that tokenizer is a mandatory argument both in documentation and code

To Reproduce
I use the python script example shared in the documentation:

from QEfficient import QEFFAutoModelForCausalLM

# Initialize the model using from_pretrained similar to transformers.AutoModelForCausalLM
model = QEFFAutoModelForCausalLM.from_pretrained("gpt2")

# Now you can directly compile the model for Cloud AI 100
model.compile(num_cores=14, device_group=[0])  # Considering you have a Cloud AI 100 Standard SKU

# You can now execute the model
model.generate(prompts=["Hi there!!"])

Expected behavior
I made the code work by making the following changes:

from QEfficient import QEFFAutoModelForCausalLM
from QEfficient.utils import load_hf_tokenizer

# Initialize the model using from_pretrained similar to transformers.AutoModelForCausalLM
model = QEFFAutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.2-1B-Instruct")

# Now you can directly compile the model for Cloud AI 100
model.compile(num_cores=16)  # Considering you have a Cloud AI 100 Standard SKU

tokenizer = load_hf_tokenizer(
        pretrained_model_name_or_path="meta-llama/Llama-3.2-1B-Instruct"
    )

# You can now execute the model
model.generate(prompts=["Hi there!!"],tokenizer=tokenizer)

Screenshots
If applicable, add screenshots to help explain your problem.

Environment (please complete the following information):

• OS: [e.g. iOS]
• Environment details with packages version etc.
• Version/Branch/Commit ID [e.g. 22]

Additional context
Add any other context about the problem here.

[quic-rishinr]

Hi @quic-shagun we recently updated the compile and generate functions but missed updating the example scripts. Thanks for pointing this out. We will get it updated shortly.

[quic-rishinr]

Hi @quic-shagun, we've updated the example scripts. Could you please review them?