mirror of
https://github.com/tloen/alpaca-lora.git
synced 2024-10-01 01:05:56 -04:00
62 lines
1.6 KiB
Python
62 lines
1.6 KiB
Python
import os
|
|
import json
|
|
|
|
import torch
|
|
from peft import PeftModel, LoraConfig
|
|
|
|
import transformers
|
|
|
|
assert (
|
|
"LlamaTokenizer" in transformers._import_structure["models.llama"]
|
|
), "LLaMA is now in HuggingFace's main branch.\nPlease reinstall it: pip uninstall transformers && pip install git+https://github.com/huggingface/transformers.git"
|
|
from transformers import LlamaTokenizer, LlamaForCausalLM
|
|
|
|
BASE_MODEL = None
|
|
assert (
|
|
BASE_MODEL
|
|
), "Please specify a BASE_MODEL in the script, e.g. 'decapoda-research/llama-7b-hf'"
|
|
|
|
tokenizer = LlamaTokenizer.from_pretrained(BASE_MODEL)
|
|
|
|
base_model = LlamaForCausalLM.from_pretrained(
|
|
BASE_MODEL,
|
|
load_in_8bit=False,
|
|
torch_dtype=torch.float16,
|
|
device_map={"": "cpu"},
|
|
)
|
|
|
|
first_weight = base_model.model.layers[0].self_attn.q_proj.weight
|
|
first_weight_old = first_weight.clone()
|
|
|
|
lora_model = PeftModel.from_pretrained(
|
|
base_model,
|
|
"tloen/alpaca-lora-7b",
|
|
device_map={"": "cpu"},
|
|
torch_dtype=torch.float16,
|
|
)
|
|
|
|
lora_weight = lora_model.base_model.model.model.layers[0].self_attn.q_proj.weight
|
|
|
|
assert torch.allclose(first_weight_old, first_weight)
|
|
|
|
# merge weights
|
|
for layer in lora_model.base_model.model.model.layers:
|
|
layer.self_attn.q_proj.merge_weights = True
|
|
layer.self_attn.v_proj.merge_weights = True
|
|
|
|
lora_model.train(False)
|
|
|
|
# did we do anything?
|
|
assert not torch.allclose(first_weight_old, first_weight)
|
|
|
|
lora_model_sd = lora_model.state_dict()
|
|
deloreanized_sd = {
|
|
k.replace("base_model.model.", ""): v
|
|
for k, v in lora_model_sd.items()
|
|
if "lora" not in k
|
|
}
|
|
|
|
LlamaForCausalLM.save_pretrained(
|
|
base_model, "./hf_ckpt", state_dict=deloreanized_sd, max_shard_size="400MB"
|
|
)
|