modules.qwen1.qwen1_configuration
Qwen1Config
Bases: EasyDeLPretrainedConfig
Source code in src/python/easydel/modules/qwen1/qwen1_configuration.py
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 |
|
add_jax_args(gradient_checkpointing='nothing_saveable', use_scan_mlp=False, scan_mlp_chunk_size=1024, bits=None, scan_layers=True, init_rope_cache_auto=False, **kwargs)
The add_jax_args function adds the following arguments to the Transformer class:
Parameters:
Name | Type | Description | Default |
---|---|---|---|
self |
Refer to the current object |
required | |
gradient_checkpointing |
str
|
str: Control the amount of memory used by jax |
'nothing_saveable'
|
use_scan_mlp |
bool
|
bool: Determine whether to use the scan_mlp function or not |
False
|
scan_mlp_chunk_size |
int
|
int: Set the chunk size for scan_mlp |
1024
|
init_rope_cache_auto |
bool
|
bool: Whether to use the rope_cache_auto in model |
False
|
bits |
Optional[int]
|
Optional[int]: Determine the number of bits used in the quantization |
None
|
scan_layers |
bool
|
bool: Determine whether to use scan layers or not |
True
|
Returns:
Type | Description |
---|---|
The following: |
Source code in src/python/easydel/modules/qwen1/qwen1_configuration.py
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 |
|
get_partition_rules(fully_sharded_data_parallel=True)
The get_partition_rules function is used to define the partitioning scheme for a model. It returns a list of tuples, where each tuple contains two elements: 1) A regex string that matches the name of one or more parameters in the model. 2) A PartitionScheme object that defines how those parameters should be partitioned across devices.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
fully_sharded_data_parallel |
bool
|
bool: Determine whether to partition the model fully or not |
True
|
Returns:
Type | Description |
---|---|
A list of tuples |
Source code in src/python/easydel/modules/qwen1/qwen1_configuration.py
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 |
|