wanyaokun
65 lines
1.9 KiB
Python
65 lines
1.9 KiB
Python
from llama_index.llms.dashscope import DashScope
|
|
from llama_index.core.base.llms.types import LLMMetadata
|
|
|
|
class DashScopeGenerationModels:
|
|
"""DashScope Qwen serial models."""
|
|
|
|
QWEN_TURBO = "qwen-turbo"
|
|
QWEN_PLUS = "qwen-plus"
|
|
QWEN_MAX = "qwen-max"
|
|
QWEN_MAX_1201 = "qwen-max-1201"
|
|
QWEN_MAX_LONGCONTEXT = "qwen-max-longcontext"
|
|
QWEN2_MATH_72B_INSTRUCT = 'qwen2-math-72b-instruct',
|
|
QWEN2_72B = 'qwen2-72b-instruct'
|
|
|
|
DASHSCOPE_MODEL_META = {
|
|
DashScopeGenerationModels.QWEN_TURBO: {
|
|
"context_window": 1024 * 8,
|
|
"num_output": 1024 * 8,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN_PLUS: {
|
|
"context_window": 1024 * 32,
|
|
"num_output": 1024 * 32,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN_MAX: {
|
|
"context_window": 1024 * 8,
|
|
"num_output": 1024 * 8,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN_MAX_1201: {
|
|
"context_window": 1024 * 8,
|
|
"num_output": 1024 * 8,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN_MAX_LONGCONTEXT: {
|
|
"context_window": 1024 * 30,
|
|
"num_output": 1024 * 30,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN2_MATH_72B_INSTRUCT: {
|
|
"context_window": 1024 * 2,
|
|
"num_output": 1024 * 8,
|
|
"is_chat_model": True,
|
|
},
|
|
DashScopeGenerationModels.QWEN2_72B: {
|
|
"context_window": 1024 * 2,
|
|
"num_output": 1024 * 8,
|
|
"is_chat_model": True,
|
|
},
|
|
}
|
|
|
|
|
|
class CustomDashScope(DashScope):
|
|
@property
|
|
def metadata(self) -> LLMMetadata:
|
|
DASHSCOPE_MODEL_META[self.model_name]["num_output"] = (
|
|
self.max_tokens or DASHSCOPE_MODEL_META[self.model_name]["num_output"]
|
|
)
|
|
return LLMMetadata(
|
|
model_name=self.model_name, **DASHSCOPE_MODEL_META[self.model_name]
|
|
)
|
|
|
|
|