tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

fix relative path for fixtures

Browse Source
This commit is contained in:
Wing Lian
2023-05-30 10:38:20 -04:00
committed by NanoCode012
parent a1f9850b91
commit cfcc549f6b
3 changed files with 18 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
src/axolotl/utils/models.py
View File

@@ -129,6 +129,7 @@ def load_model(
llm_int8_threshold=6.0, llm_int8_threshold=6.0,
llm_int8_has_fp16_weight=False, llm_int8_has_fp16_weight=False,
bnb_4bit_compute_dtype=torch_dtype, bnb_4bit_compute_dtype=torch_dtype,
bnb_4bit_compute_dtype=torch_dtype,
bnb_4bit_use_double_quant=True, bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type="nf4", bnb_4bit_quant_type="nf4",
) )
@@ -280,8 +281,8 @@ def load_model(
# llama is PROBABLY model parallelizable, but the default isn't that it is # llama is PROBABLY model parallelizable, but the default isn't that it is
# so let's only set it for the 4bit, see # so let's only set it for the 4bit, see
# https://github.com/johnsmith0031/alpaca_lora_4bit/blob/08b3fca4a4a9e0d3945be1bab4529f100a428636/finetune.py#L130-L133 # https://github.com/johnsmith0031/alpaca_lora_4bit/blob/08b3fca4a4a9e0d3945be1bab4529f100a428636/finetune.py#L130-L133
setattr(model, 'is_parallelizable', True) setattr(model, "is_parallelizable", True)
setattr(model, 'model_parallel', True) setattr(model, "model_parallel", True)
requires_grad = [] requires_grad = []
for name, param in model.named_parameters(recurse=True): for name, param in model.named_parameters(recurse=True):

3
src/axolotl/utils/trainer.py
View File

@@ -125,7 +125,8 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
output_dir=cfg.output_dir, output_dir=cfg.output_dir,
save_total_limit=3, save_total_limit=3,
load_best_model_at_end=( load_best_model_at_end=(
cfg.val_set_size > 0 cfg.load_best_model_at_end is not False
and cfg.val_set_size > 0
and save_steps and save_steps
and save_steps % eval_steps == 0 and save_steps % eval_steps == 0
and cfg.load_in_8bit is not True and cfg.load_in_8bit is not True

15
tests/test_prompt_tokenizers.py
View File

@@ -1,6 +1,8 @@
"""Module for testing prompt tokenizers."""
import json import json
import logging import logging
import unittest import unittest
from pathlib import Path from pathlib import Path
from transformers import AutoTokenizer from transformers import AutoTokenizer
@@ -12,6 +14,10 @@ logging.basicConfig(level="INFO")
class TestPromptTokenizationStrategies(unittest.TestCase): class TestPromptTokenizationStrategies(unittest.TestCase):
"""
Test class for prompt tokenization strategies.
"""
def setUp(self) -> None: def setUp(self) -> None:
self.tokenizer = AutoTokenizer.from_pretrained("huggyllama/llama-7b") self.tokenizer = AutoTokenizer.from_pretrained("huggyllama/llama-7b")
self.tokenizer.add_special_tokens( self.tokenizer.add_special_tokens(
@@ -24,10 +30,15 @@ class TestPromptTokenizationStrategies(unittest.TestCase):
def test_sharegpt_integration(self): def test_sharegpt_integration(self):
print(Path(__file__).parent) print(Path(__file__).parent)
with open(Path(__file__).parent / "fixtures/conversation.json", "r") as fin: with open(
Path(__file__).parent / "fixtures/conversation.json", encoding="utf-8"
) as fin:
data = fin.read() data = fin.read()
conversation = json.loads(data) conversation = json.loads(data)
with open(Path(__file__).parent / "fixtures/conversation.tokenized.json", "r") as fin: with open(
Path(__file__).parent / "fixtures/conversation.tokenized.json",
encoding="utf-8",
) as fin:
data = fin.read() data = fin.read()
tokenized_conversation = json.loads(data) tokenized_conversation = json.loads(data)
prompter = ShareGPTPrompter("chat") prompter = ShareGPTPrompter("chat")