qubee.configs.llm_config.LlmConfig Class Reference#
|
Mobilint SDK qb Compiler v0.11.0.1
MCS002-EN
|
qubee.configs.llm_config.LlmConfig Class Reference
Inheritance diagram for qubee.configs.llm_config.LlmConfig:
Public Member Functions | |
| __init__ (self, int max_data_len, int max_seq_len, int max_cache_len, int random_seq_len_calib, bool use_full_seq_len_calib, bool use_custom_mask_input, int start_dram_offset, bool use_global_core, int llm_batch_size, list[int] npu_core_ids, bool optional=True) | |
| default_config (cls) | |
| to_dict (self) | |
Public Attributes | |
| max_data_len = max_data_len | |
| max_seq_len = max_seq_len | |
| max_cache_len = max_cache_len | |
| random_seq_len_calib = random_seq_len_calib | |
| use_full_seq_len_calib = use_full_seq_len_calib | |
| use_custom_mask_input = use_custom_mask_input | |
| start_dram_offset = start_dram_offset | |
| use_global_core = use_global_core | |
| llm_batch_size = llm_batch_size | |
| npu_core_ids = npu_core_ids | |
Static Public Attributes | |
| dict | py_to_cpp_name |
| dict | min_max_check |
Detailed Description
Definition at line 21 of file llm_config.py.
Constructor & Destructor Documentation
◆ __init__()
| qubee.configs.llm_config.LlmConfig.__init__ | ( | self, | |
| int | max_data_len, | ||
| int | max_seq_len, | ||
| int | max_cache_len, | ||
| int | random_seq_len_calib, | ||
| bool | use_full_seq_len_calib, | ||
| bool | use_custom_mask_input, | ||
| int | start_dram_offset, | ||
| bool | use_global_core, | ||
| int | llm_batch_size, | ||
| list[int] | npu_core_ids, | ||
| bool | optional = True ) |
Definition at line 48 of file llm_config.py.
Member Function Documentation
◆ default_config()
| qubee.configs.llm_config.LlmConfig.default_config | ( | cls | ) |
Definition at line 76 of file llm_config.py.
◆ to_dict()
| qubee.configs.llm_config.LlmConfig.to_dict | ( | self | ) |
Definition at line 102 of file llm_config.py.
Member Data Documentation
◆ py_to_cpp_name
|
static |
Initial value:
= {
"max_data_len": "maxDataLen",
"max_seq_len": "maxSeqLen",
"max_cache_len": "maxCacheLen",
"random_seq_len_calib": "randomSeqLenCalib",
"use_full_seq_len_calib": "useFullSeqLenCalib",
"use_custom_mask_input": "useCustomMaskInput",
"start_dram_offset": "startDramOffset",
"use_global_core": "useGlobalCore",
"llm_batch_size": "llmBatchSize",
"npu_core_ids": "npuCoreIds",
}
Definition at line 22 of file llm_config.py.
◆ min_max_check
|
static |
Initial value:
= {
"max_data_len": [0, None],
"max_seq_len": [0, None],
"max_cache_len": [0, None],
"random_seq_len_calib": [0, None],
"use_full_seq_len_calib": [0, 1],
"use_custom_mask_input": [0, 1],
"start_dram_offset": [0, None],
"use_global_core": [0, 1],
"llm_batch_size": [0, None],
}
Definition at line 35 of file llm_config.py.
◆ max_data_len
| qubee.configs.llm_config.LlmConfig.max_data_len = max_data_len |
Definition at line 63 of file llm_config.py.
◆ max_seq_len
| qubee.configs.llm_config.LlmConfig.max_seq_len = max_seq_len |
Definition at line 64 of file llm_config.py.
◆ max_cache_len
| qubee.configs.llm_config.LlmConfig.max_cache_len = max_cache_len |
Definition at line 65 of file llm_config.py.
◆ random_seq_len_calib
| qubee.configs.llm_config.LlmConfig.random_seq_len_calib = random_seq_len_calib |
Definition at line 66 of file llm_config.py.
◆ use_full_seq_len_calib
| qubee.configs.llm_config.LlmConfig.use_full_seq_len_calib = use_full_seq_len_calib |
Definition at line 67 of file llm_config.py.
◆ use_custom_mask_input
| qubee.configs.llm_config.LlmConfig.use_custom_mask_input = use_custom_mask_input |
Definition at line 68 of file llm_config.py.
◆ start_dram_offset
| qubee.configs.llm_config.LlmConfig.start_dram_offset = start_dram_offset |
Definition at line 69 of file llm_config.py.
◆ use_global_core
| qubee.configs.llm_config.LlmConfig.use_global_core = use_global_core |
Definition at line 70 of file llm_config.py.
◆ llm_batch_size
| qubee.configs.llm_config.LlmConfig.llm_batch_size = llm_batch_size |
Definition at line 71 of file llm_config.py.
◆ npu_core_ids
| qubee.configs.llm_config.LlmConfig.npu_core_ids = npu_core_ids |
Definition at line 72 of file llm_config.py.
The documentation for this class was generated from the following file:
Generated by