tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

simplify by removing duplicate base_model_config (#772)

Browse Source
This commit is contained in:
Wing Lian
2023-10-23 01:42:38 -04:00
committed by GitHub
parent 44c9d0151a
commit 2d8def68dc
39 changed files with 10 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
tests/e2e/test_fused_llama.py
View File

@@ -31,7 +31,6 @@ class TestFusedLlama(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "JackFram/llama-68m",
"base_model_config": "JackFram/llama-68m",
"flash_attention": True,
"flash_attn_fuse_qkv": True,
"flash_attn_fuse_mlp": True,

3
tests/e2e/test_lora_llama.py
View File

@@ -29,7 +29,6 @@ class TestLoraLlama(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "JackFram/llama-68m",
"base_model_config": "JackFram/llama-68m",
"tokenizer_type": "LlamaTokenizer",
"sequence_len": 1024,
"load_in_8bit": True,
@@ -72,7 +71,6 @@ class TestLoraLlama(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "JackFram/llama-68m",
"base_model_config": "JackFram/llama-68m",
"tokenizer_type": "LlamaTokenizer",
"sequence_len": 1024,
"sample_packing": True,
@@ -117,7 +115,6 @@ class TestLoraLlama(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "TheBlokeAI/jackfram_llama-68m-GPTQ",
"base_model_config": "TheBlokeAI/jackfram_llama-68m-GPTQ",
"model_type": "AutoModelForCausalLM",
"tokenizer_type": "LlamaTokenizer",
"sequence_len": 1024,

2
tests/e2e/test_mistral.py
View File

@@ -31,7 +31,6 @@ class TestMistral(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "openaccess-ai-collective/tiny-mistral",
"base_model_config": "openaccess-ai-collective/tiny-mistral",
"flash_attention": True,
"sequence_len": 1024,
"load_in_8bit": True,
@@ -77,7 +76,6 @@ class TestMistral(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "openaccess-ai-collective/tiny-mistral",
"base_model_config": "openaccess-ai-collective/tiny-mistral",
"flash_attention": True,
"sequence_len": 1024,
"val_set_size": 0.1,

2
tests/e2e/test_mistral_samplepack.py
View File

@@ -31,7 +31,6 @@ class TestMistral(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "openaccess-ai-collective/tiny-mistral",
"base_model_config": "openaccess-ai-collective/tiny-mistral",
"flash_attention": True,
"sample_packing": True,
"sequence_len": 1024,
@@ -78,7 +77,6 @@ class TestMistral(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "openaccess-ai-collective/tiny-mistral",
"base_model_config": "openaccess-ai-collective/tiny-mistral",
"flash_attention": True,
"sample_packing": True,
"sequence_len": 1024,

2
tests/e2e/test_phi.py
View File

@@ -27,7 +27,6 @@ class TestPhi(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "microsoft/phi-1_5",
"base_model_config": "microsoft/phi-1_5",
"trust_remote_code": True,
"model_type": "MixFormerSequentialForCausalLM",
"tokenizer_type": "AutoTokenizer",
@@ -71,7 +70,6 @@ class TestPhi(unittest.TestCase):
cfg = DictDefault(
{
"base_model": "microsoft/phi-1_5",
"base_model_config": "microsoft/phi-1_5",
"trust_remote_code": True,
"model_type": "MixFormerSequentialForCausalLM",
"tokenizer_type": "AutoTokenizer",

7
tests/test_normalize_config.py
View File

@@ -37,3 +37,10 @@ class NormalizeConfigTestCase(unittest.TestCase):
normalize_config(cfg)
assert cfg.learning_rate == 0.00005
def test_base_model_config_set_when_empty(self):
cfg = self._get_base_cfg()
del cfg.base_model_config
normalize_config(cfg)
assert cfg.base_model_config == cfg.base_model