apunkt/doc-to-lora
1
0
Fork 0
mirror of https://github.com/SakanaAI/doc-to-lora.git synced 2026-07-23 17:01:04 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

icml rebuttal

Browse source
This commit is contained in:
51616 2026-06-15 04:31:47 +00:00
parent 22267c7666
commit 696b45c51b
44 changed files with 5300 additions and 40 deletions
Download patch file Download diff file Expand all files Collapse all files

19
examples/python_api.py
View file

@ -5,7 +5,7 @@ import torch
from ctx_to_lora.model_loading import get_tokenizer
from ctx_to_lora.modeling.hypernet import ModulatedPretrainedModel
checkpoint_path = "../../ctx-to-lora/train_outputs/runs/Sep29_14-42-46_slurm0-a3nodeset-9_88483_1e7bb34e/checkpoint-40000/pytorch_model.bin"
checkpoint_path = "trained_d2l/gemma_demo/checkpoint-80000/pytorch_model.bin"
state_dict = torch.load(checkpoint_path, weights_only=False)
model = ModulatedPretrainedModel.from_state_dict(
state_dict, train=False, use_sequence_packing=False
@ -14,8 +14,13 @@ model.reset()
tokenizer = get_tokenizer(model.base_model.name_or_path)
doc = open("data/sakana_wiki.txt").read()
chat = [{"role": "user", "content": "Summarize what Sakana AI does."}]
doc = open("data/bitter_lesson.txt").read()
chat = [
{
"role": "user",
"content": "What is the bitter lesson in AI research?",
}
]
chat_ids = tokenizer.apply_chat_template(
chat,
add_special_tokens=False,
@ -36,8 +41,8 @@ outputs = model.generate(input_ids=chat_ids, max_new_tokens=256)
print(tokenizer.decode(outputs[0]))
# remove internalized info
model.reset()
# # remove internalized info
# model.reset()
outputs = model.generate(input_ids=chat_ids, max_new_tokens=256)
print(tokenizer.decode(outputs[0]))
# outputs = model.generate(input_ids=chat_ids, max_new_tokens=256)
# print(tokenizer.decode(outputs[0]))