File size: 5,920 Bytes
7c2bf30 cc11c6b 7c2bf30 cc11c6b 7c2bf30 cc11c6b 7c2bf30 cc11c6b 7c2bf30 |
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 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 |
"""
tests for chat_template prompt strategy
"""
import unittest
import pytest
from datasets import Dataset
from transformers import AutoTokenizer
from axolotl.prompt_strategies.chat_template import (
ChatTemplatePrompter,
ChatTemplateStrategy,
load,
)
from axolotl.utils.chat_templates import chat_templates
from axolotl.utils.dict import DictDefault
@pytest.fixture(name="assistant_dataset")
def fixture_assistant_dataset():
# pylint: disable=duplicate-code
return Dataset.from_list(
[
{
"messages": [
{
"role": "user",
"content": "hello",
},
{
"role": "assistant",
"content": "hello",
},
{
"role": "user",
"content": "goodbye",
},
{
"role": "assistant",
"content": "goodbye",
},
]
}
]
)
@pytest.fixture(name="sharegpt_dataset")
def fixture_sharegpt_dataset():
# pylint: disable=duplicate-code
return Dataset.from_list(
[
{
"conversations": [
{
"from": "human",
"value": "hello",
},
{
"from": "gpt",
"value": "hello",
},
{
"from": "human",
"value": "goodbye",
},
{
"from": "gpt",
"value": "goodbye",
},
]
}
]
)
@pytest.fixture(name="llama3_tokenizer")
def fixture_llama3_tokenizer():
tokenizer = AutoTokenizer.from_pretrained("NousResearch/Meta-Llama-3-8B")
tokenizer.eos_token = "<|eot_id|>"
return tokenizer
class TestAssistantChatTemplateLlama3:
"""
Test class for assistant style datasets with llama-3 prompts using the chat_template strategy.
"""
def test_llama3_load(self, llama3_tokenizer, assistant_dataset):
# pylint: disable=duplicate-code
strategy = load(
llama3_tokenizer,
DictDefault(
{
"train_on_inputs": False,
"sequence_len": 512,
}
),
DictDefault(
{
"chat_template": "llama3",
"message_field_role": "role",
"message_field_content": "content",
"roles": {
"user": ["user"],
"assistant": ["assistant"],
"system": ["system"],
},
"field_messages": "messages",
}
),
)
res = strategy.tokenize_prompt(assistant_dataset[0])
input_ids = res["input_ids"]
# fmt: off
assert input_ids == [
128000, # bos
128006, 882, 128007, # user header
271, 15339, 128009, # user prompt eot
128006, 78191, 128007, # assistant header
271, 15339, 128009, # assistant response eot
128006, 882, 128007,
271, 19045, 29474, 128009,
128006, 78191, 128007,
271, 19045, 29474, 128009,
]
# fmt: on
def test_llama3(self, llama3_tokenizer, assistant_dataset):
# pylint: disable=duplicate-code
strategy = ChatTemplateStrategy(
ChatTemplatePrompter(
llama3_tokenizer,
chat_templates("llama3"),
message_field_role="role",
message_field_content="content",
roles={
"user": ["user"],
"assistant": ["assistant"],
"system": ["system"],
},
),
llama3_tokenizer,
False,
512,
)
strategy.messages = "messages"
res = strategy.tokenize_prompt(assistant_dataset[0])
input_ids = res["input_ids"]
# fmt: off
assert input_ids == [
128000, # bos
128006, 882, 128007, # user header
271, 15339, 128009, # user prompt eot
128006, 78191, 128007, # assistant header
271, 15339, 128009, # assistant response eot
128006, 882, 128007,
271, 19045, 29474, 128009,
128006, 78191, 128007,
271, 19045, 29474, 128009,
]
# fmt: on
class TestSharegptChatTemplateLlama3:
"""
Test class for ShareGPT style datasets with llama-3 prompts using the chat_template strategy.
"""
def test_llama3(self, llama3_tokenizer, sharegpt_dataset):
# pylint: disable=duplicate-code
strategy = ChatTemplateStrategy(
ChatTemplatePrompter(llama3_tokenizer, chat_templates("llama3")),
llama3_tokenizer,
False,
512,
)
res = strategy.tokenize_prompt(sharegpt_dataset[0])
input_ids = res["input_ids"]
# fmt: off
assert input_ids == [
128000, # bos
128006, 882, 128007, # user header
271, 15339, 128009, # user prompt eot
128006, 78191, 128007, # assistant header
271, 15339, 128009, # assistant response eot
128006, 882, 128007,
271, 19045, 29474, 128009,
128006, 78191, 128007,
271, 19045, 29474, 128009,
]
# fmt: on
if __name__ == "__main__":
unittest.main()
|