github
/
ColossalAI
mirror of https://github.com/hpcaitech/ColossalAI
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[CI/tests] simplify some test case to reduce testing time (#5755) 

* [ci/tests] simplify some test case to reduce testing time

* [ci/tests] continue to remove test case to reduce ci time cost

* restore some test config

* [ci/tests] continue to reduce ci time cost
pull/5753/head
Haze188 2024-06-04 13:57:54 +08:00 committed by GitHub
parent 406443200f
commit e22b82755d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 5 additions and 167 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
tests/test_shardformer/test_model/test_shard_bert.py
View File

@ -122,20 +122,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 2,
"pp_size": 1,
"enable_all_optimization": True,
"use_lazy_init": True,
"precision": "fp32",
},
{
"tp_size": 1,
"pp_size": 2,
"num_microbatches": 4,
"use_lazy_init": True,
"precision": "fp32",
},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 2, "pp_size": 2,
@ -145,14 +131,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 4,
"pp_size": 1,
"enable_all_optimization": True,
"use_lazy_init": False,
"precision": "fp32",
},
{"tp_size": 2, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 1, "pp_size": 1,
@ -162,16 +140,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 2,
"num_microbatches": 2,
"enable_all_optimization": True,
"use_lazy_init": True,
"zero_stage": 1,
"precision": "fp16",
"initial_scale": 1,
},
], ],
) )
def run_bert_test(test_config): def run_bert_test(test_config):

4
tests/test_shardformer/test_model/test_shard_blip2.py
View File

@ -67,8 +67,8 @@ def check_forward_backward(org_model, sharded_model, data_gen_fn, output_transfo
@parameterize("enable_fused_normalization", [True, False]) @parameterize("enable_fused_normalization", [True, False])
@parameterize("enable_tensor_parallelism", [True, False]) @parameterize("enable_tensor_parallelism", [True, False])
@parameterize("enable_flash_attention", [True, False]) @parameterize("enable_flash_attention", [True])
@parameterize("enable_jit_fused", [True, False]) @parameterize("enable_jit_fused", [True])
def run_blip2_test( def run_blip2_test(
enable_fused_normalization, enable_fused_normalization,
enable_tensor_parallelism, enable_tensor_parallelism,

23
tests/test_shardformer/test_model/test_shard_bloom.py
View File

@ -110,17 +110,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp32", "precision": "fp32",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 4,
"pp_size": 1,
"num_microbatches": 1,
"enable_sequence_parallelism": True,
"sequence_parallelism_mode": "split_gather",
"enable_flash_attention": False,
"use_lazy_init": True,
"precision": "fp16",
"initial_scale": 1,
},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 2, "pp_size": 2,
@ -128,6 +117,7 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"enable_all_optimization": True, "enable_all_optimization": True,
"use_lazy_init": True, "use_lazy_init": True,
"precision": "fp16", "precision": "fp16",
"zero_stage": 1,
"initial_scale": 1, "initial_scale": 1,
}, },
{ {
@ -138,17 +128,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"use_lazy_init": False, "use_lazy_init": False,
"precision": "fp32", "precision": "fp32",
}, },
{"tp_size": 4, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{"tp_size": 2, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{
"tp_size": 2,
"pp_size": 1,
"enable_all_optimization": True,
"use_lazy_init": True,
"zero_stage": 2,
"precision": "fp16",
"initial_scale": 1,
},
{ {
"tp_size": 1, "tp_size": 1,
"pp_size": 2, "pp_size": 2,

23
tests/test_shardformer/test_model/test_shard_falcon.py
View File

@ -92,21 +92,12 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"tp_size": 2, "tp_size": 2,
"pp_size": 2, "pp_size": 2,
"num_microbatches": 4, "num_microbatches": 4,
"enable_all_optimization": True, "enable_all_optimization": False,
"use_lazy_init": True, "use_lazy_init": False,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 2,
"num_microbatches": 4,
"enable_all_optimization": False,
"use_lazy_init": False,
"precision": "fp32",
},
{"tp_size": 4, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"}, {"tp_size": 4, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{"tp_size": 2, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 1, "pp_size": 1,
@ -116,16 +107,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 2,
"num_microbatches": 2,
"enable_all_optimization": True,
"use_lazy_init": True,
"zero_stage": 1,
"precision": "fp16",
"initial_scale": 1,
},
], ],
) )
def run_falcon_test(test_config): def run_falcon_test(test_config):

40
tests/test_shardformer/test_model/test_shard_gpt2.py
View File

@ -162,46 +162,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 2,
"num_microbatches": 4,
"enable_all_optimization": True,
"use_lazy_init": True,
"precision": "fp16",
"initial_scale": 1,
},
{
"tp_size": 4,
"pp_size": 1,
"enable_all_optimization": False,
"use_lazy_init": False,
"precision": "fp32",
},
{
"tp_size": 2,
"pp_size": 1,
"enable_all_optimization": False,
"use_lazy_init": False,
"precision": "fp32",
},
{
"tp_size": 2,
"pp_size": 2,
"num_microbatches": 4,
"enable_all_optimization": False,
"use_lazy_init": True,
"precision": "fp32",
},
{
"tp_size": 2,
"pp_size": 1,
"enable_all_optimization": True,
"use_lazy_init": True,
"zero_stage": 2,
"precision": "fp16",
"initial_scale": 1,
},
{ {
"tp_size": 1, "tp_size": 1,
"pp_size": 2, "pp_size": 2,

49
tests/test_shardformer/test_model/test_shard_llama.py
View File

@ -145,17 +145,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 4,
"pp_size": 1,
"num_microbatches": 1,
"enable_sequence_parallelism": True,
"sequence_parallelism_mode": "ring",
"enable_flash_attention": False,
"use_lazy_init": True,
"precision": "fp32",
"initial_scale": 1,
},
{ {
"tp_size": 4, "tp_size": 4,
"pp_size": 1, "pp_size": 1,
@ -167,17 +156,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 1,
"sp_size": 2,
"num_microbatches": 1,
"enable_sequence_parallelism": True,
"sequence_parallelism_mode": "all_to_all",
"use_lazy_init": True,
"precision": "fp16",
"initial_scale": 1,
},
{ {
"tp_size": 1, "tp_size": 1,
"pp_size": 1, "pp_size": 1,
@ -190,17 +168,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp16", "precision": "fp16",
"initial_scale": 1, "initial_scale": 1,
}, },
{
"tp_size": 1,
"pp_size": 1,
"num_microbatches": 1,
"enable_sequence_parallelism": True,
"sequence_parallelism_mode": "all_to_all",
"enable_flash_attention": False,
"use_lazy_init": True,
"precision": "fp16",
"initial_scale": 1,
},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 2, "pp_size": 2,
@ -221,22 +188,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"enable_gradient_checkpointing": True, "enable_gradient_checkpointing": True,
"gradient_checkpoint_config": PipelineGradientCheckpointConfig(num_ckpt_layers_per_stage=[4, 0]), "gradient_checkpoint_config": PipelineGradientCheckpointConfig(num_ckpt_layers_per_stage=[4, 0]),
}, },
{
"tp_size": 4,
"pp_size": 1,
"enable_all_optimization": False,
"use_lazy_init": False,
"precision": "fp32",
},
{
"tp_size": 1,
"pp_size": 4,
"num_microbatches": 4,
"enable_all_optimization": False,
"use_lazy_init": False,
"precision": "fp32",
},
{"tp_size": 2, "pp_size": 1, "enable_all_optimization": False, "use_lazy_init": False, "precision": "fp32"},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 1, "pp_size": 1,

1
tests/test_shardformer/test_model/test_shard_vit.py
View File

@ -108,7 +108,6 @@ def check_forward_backward(model_fn, data_gen_fn, output_transform_fn, loss_fn,
"precision": "fp32", "precision": "fp32",
}, },
{"tp_size": 4, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"}, {"tp_size": 4, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{"tp_size": 2, "pp_size": 1, "enable_all_optimization": True, "use_lazy_init": False, "precision": "fp32"},
{ {
"tp_size": 2, "tp_size": 2,
"pp_size": 1, "pp_size": 1,