init
This commit is contained in:
52
MemoAI/qwen2-0.5b/di_config.json
Normal file
52
MemoAI/qwen2-0.5b/di_config.json
Normal file
@@ -0,0 +1,52 @@
|
||||
{
|
||||
"model_space": "YOUR-NAME-SPACE",
|
||||
"model_name": "YOUR-MODEL-NAME",
|
||||
"model_type": "Qwen_v20",
|
||||
"model_path": "./dashinfer_models/",
|
||||
"data_type": "float32",
|
||||
"device_type": "CPU",
|
||||
"device_ids": [
|
||||
0
|
||||
],
|
||||
"multinode_mode": false,
|
||||
"engine_config": {
|
||||
"engine_max_length": 1024,
|
||||
"engine_max_batch": 2,
|
||||
"do_profiling": false,
|
||||
"num_threads": 0,
|
||||
"matmul_precision": "medium"
|
||||
},
|
||||
"generation_config": {
|
||||
"temperature": 0.7,
|
||||
"early_stopping": true,
|
||||
"top_k": 20,
|
||||
"top_p": 0.8,
|
||||
"repetition_penalty": 1.05,
|
||||
"presence_penalty": 0.0,
|
||||
"min_length": 0,
|
||||
"max_length": 8192,
|
||||
"no_repeat_ngram_size": 0,
|
||||
"eos_token_id": 151643,
|
||||
"seed": 1234,
|
||||
"stop_words_ids": [
|
||||
[
|
||||
151643
|
||||
],
|
||||
[
|
||||
151644
|
||||
],
|
||||
[
|
||||
151645
|
||||
]
|
||||
]
|
||||
},
|
||||
"convert_config": {
|
||||
"do_dynamic_quantize_convert": false
|
||||
},
|
||||
"quantization_config": {
|
||||
"activation_type": "bfloat16",
|
||||
"weight_type": "uint8",
|
||||
"SubChannel": true,
|
||||
"GroupSize": 512
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user