# qairt.modules.genie\_execution ## qairt.modules.genie\_execution.genie\_config - *class* qairt.modules.genie\_execution.genie\_config.AbstractDialog(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - context*: [Context](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.Context)* - - embedding*: Optional[[DialogEmbedding](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEmbedding)]* *= None* - - max\_num\_tokens*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='max-num-tokens', alias\_priority=2)* - - sampler*: Optional[[Sampler](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.Sampler)]* *= None* - - stop\_sequence*: Optional[List[str]]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='stop-sequence', alias\_priority=2)* - - tokenizer*: [Tokenizer](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.Tokenizer)* - - type*: str* - - *class* qairt.modules.genie\_execution.genie\_config.BasicDialog(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`AbstractDialog`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.AbstractDialog) - engine*: [DialogEngine](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEngine)* - - type*: Literal['basic']* *= 'basic'* - - *class* qairt.modules.genie\_execution.genie\_config.Context(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - bos\_token*: int* *= FieldInfo(annotation=NoneType, required=False, default=0, alias='bos-token', alias\_priority=2)* - - eos\_token*: Union[int, List[int]]* *= FieldInfo(annotation=NoneType, required=False, default=0, alias='eos-token', alias\_priority=2)* - - eot\_token*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='eot-token', alias\_priority=2)* - - n\_embd*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-embd', alias\_priority=2)* - - n\_vocab*: int* *= FieldInfo(annotation=NoneType, required=False, default=0, alias='n-vocab', alias\_priority=2)* - - pad\_token*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='pad-token', alias\_priority=2)* - - size*: int* *= 512* - - *class* qairt.modules.genie\_execution.genie\_config.DialogEmbedding(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - datatype*: [DialogEmbeddingDataType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEmbeddingDataType)* - - lut\_path*: str | os.PathLike* *= FieldInfo(annotation=NoneType, required=True, alias='lut-path', alias\_priority=2)* - - quant\_param*: Optional[[DialogEmbeddingQuantParam](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEmbeddingQuantParam)]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='quant-param', alias\_priority=2)* - - size*: int* - - type*: str* *= 'lut'* - - *class* qairt.modules.genie\_execution.genie\_config.DialogEmbeddingDataType(*value*) - Bases: `str`, `Enum` An enumeration. - FLOAT32 *= 'float32'* - - NATIVE *= 'native'* - - UFIXED16 *= 'ufixed16'* - - UFIXED4 *= 'ufixed4'* - - UFIXED8 *= 'ufixed8'* - - *class* qairt.modules.genie\_execution.genie\_config.DialogEmbeddingQuantParam(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - offset*: int* - - scale*: float* - - *class* qairt.modules.genie\_execution.genie\_config.DialogEngine(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - backend*: [EngineBackend](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EngineBackend)* *= FieldInfo(annotation=NoneType, required=False, default\_factory=EngineBackend)* - - model*: [EngineModel](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EngineModel)* *= FieldInfo(annotation=NoneType, required=False, default\_factory=EngineModel)* - - n\_threads*: int* *= FieldInfo(annotation=NoneType, required=False, default=6, alias='n-threads', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.DialogType(*value*) - Bases: `str`, `Enum` An enumeration. - BASIC *= 'basic'* - - EAGLET *= 'eaglet'* - - LADE *= 'lade'* - - SSD\_Q1 *= 'ssd-q1'* - - *class* qairt.modules.genie\_execution.genie\_config.DraftEngineModel(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`EngineModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EngineModel) - draft\_token\_map*: Optional[Union[str, PathLike]]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='draft-token-map', alias\_priority=2)* - - parse\_draft\_token\_map(*v*) - - *class* qairt.modules.genie\_execution.genie\_config.EagletConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - depth*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='depth', alias\_priority=2)* - - draft\_kv\_cache*: bool* *= FieldInfo(annotation=NoneType, required=True, alias='draft-kv-cache', alias\_priority=2)* - - draft\_len*: int* *= FieldInfo(annotation=NoneType, required=True, alias='draft-len', alias\_priority=2)* - - eaglet\_version*: Union[int, str]* *= FieldInfo(annotation=NoneType, required=False, default=1, alias='eaglet-version', alias\_priority=2)* - - gdn\_chunk\_size*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='gdn-chunk-size', alias\_priority=2)* - - max\_tokens\_target\_can\_evaluate*: int* *= FieldInfo(annotation=NoneType, required=True, alias='max-tokens-target-can-evaluate', alias\_priority=2)* - - n\_branches*: int* *= FieldInfo(annotation=NoneType, required=True, alias='n-branches', alias\_priority=2)* - - prepend\_accepted\_tokens*: Optional[bool]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='prepend-accepted-tokens', alias\_priority=2)* - - supported\_draft\_shapes*: Optional[List[str]]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='supported-draft-shapes', alias\_priority=2)* - - top\_k*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='top-k', alias\_priority=2)* - - total\_tokens*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='total-tokens', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.EagletContext(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`Context`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.Context) - draft\_n\_vocab*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='draft-n-vocab', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.EagletDialog(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`AbstractDialog`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.AbstractDialog) - context*: [EagletContext](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EagletContext)* - - eaglet*: [EagletConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EagletConfig)* - - engine*: List[[qairt.modules.genie\_execution.genie\_config.EagletDraftDialogEngine](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EagletDraftDialogEngine) | [qairt.modules.genie\_execution.genie\_config.EagletTargetDialogEngine](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EagletTargetDialogEngine)]* - - type*: Literal['eaglet']* *= 'eaglet'* - - *class* qairt.modules.genie\_execution.genie\_config.EagletDraftDialogEngine(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`DialogEngine`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEngine) - model*: [DraftEngineModel](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DraftEngineModel)* *= FieldInfo(annotation=NoneType, required=False, default\_factory=DraftEngineModel)* - - role*: DRAFT: 'draft'>]* *= 'draft'* - - *class* qairt.modules.genie\_execution.genie\_config.EagletEngineRole(*value*) - Bases: `str`, `Enum` An enumeration. - DRAFT *= 'draft'* - - TARGET *= 'target'* - - *class* qairt.modules.genie\_execution.genie\_config.EagletTargetDialogEngine(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`DialogEngine`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEngine) - role*: TARGET: 'target'>]* *= 'target'* - - *class* qairt.modules.genie\_execution.genie\_config.EngineBackend(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - QnnGenAiTransformer*: Optional[[QnnGenAiTransformerBackend](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.QnnGenAiTransformerBackend)]* *= None* - - QnnHtp*: Optional[[QnnHtpBackend](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.QnnHtpBackend)]* *= None* - - check\_type() → Self - - extensions*: Optional[Union[str, PathLike]]* *= None* - - type*: [EngineBackendType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EngineBackendType)* *= 'QnnGenAiTransformer'* - - *class* qairt.modules.genie\_execution.genie\_config.EngineBackendType(*value*) - Bases: `str`, `Enum` An enumeration. - QNN\_GEN\_AI\_TRANSFORMER *= 'QnnGenAiTransformer'* - - QNN\_HTP *= 'QnnHtp'* - - *class* qairt.modules.genie\_execution.genie\_config.EngineModel(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - binary*: Optional[[ModelBinary](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.ModelBinary)]* *= None* - - library*: Optional[[ModelLibrary](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.ModelLibrary)]* *= None* - - linear\_attention*: Optional[bool]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='linear-attention', alias\_priority=2)* - - positional\_encoding*: Optional[[PositionalEncoding](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.PositionalEncoding)]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='positional-encoding', alias\_priority=2)* - - type*: [EngineModelType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EngineModelType)* *= 'library'* - - *class* qairt.modules.genie\_execution.genie\_config.EngineModelType(*value*) - Bases: `str`, `Enum` An enumeration. - BINARY *= 'binary'* - - LIBRARY *= 'library'* - - *class* qairt.modules.genie\_execution.genie\_config.ExportFormat(*value*) - Bases: `Enum` Container export formats - DIALOG *= 'dialog'* - - LM\_EXECUTOR *= 'lm\_executor'* - - *class* qairt.modules.genie\_execution.genie\_config.GenieConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: `AISWBaseModel` top level config object for genie config - dialog*: Union[[BasicDialog](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.BasicDialog), [LadeDialog](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LadeDialog), [SSDDialog](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.SSDDialog), [EagletDialog](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.EagletDialog)]* *= FieldInfo(annotation=NoneType, required=True, discriminator='type')* - - export(*export\_format: [ExportFormat](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.ExportFormat) = ExportFormat.DIALOG*) → dict[str, Any] - - *class* qairt.modules.genie\_execution.genie\_config.GenieConfigEncoder(*\**, *skipkeys=False*, *ensure\_ascii=True*, *check\_circular=True*, *allow\_nan=True*, *sort\_keys=False*, *indent=None*, *separators=None*, *default=None*) - Bases: `JSONEncoder` - default(*obj*) - Implement this method in a subclass such that it returns a serializable object for `o`, or calls the base implementation (to raise a `TypeError`). For example, to support arbitrary iterators, you could implement default like this: def default(self, o): try: iterable = iter(o) except TypeError: pass else: return list(iterable) # Let the base class default method raise the TypeError return JSONEncoder.default(self, o) Copy to clipboard - *class* qairt.modules.genie\_execution.genie\_config.LADEType(*value*) - Bases: `str`, `Enum` An enumeration. - ALWAYS\_FWD\_ONE *= 'ALWAYS\_FWD\_ONE'* - - FWD\_LEVEL *= 'FWD\_LEVEL'* - - FWD\_MAX\_HIT *= 'FWD\_MAX\_HIT'* - - *class* qairt.modules.genie\_execution.genie\_config.LadeConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - gcap*: int* - - ngram*: int* - - update\_mode*: [LADEType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LADEType)* *= FieldInfo(annotation=NoneType, required=False, default=<LADEType.ALWAYS\_FWD\_ONE: 'ALWAYS\_FWD\_ONE'>, alias='update-mode', alias\_priority=2)* - - window*: int* - - *class* qairt.modules.genie\_execution.genie\_config.LadeDialog(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`AbstractDialog`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.AbstractDialog) - engine*: [DialogEngine](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEngine)* - - lade*: [LadeConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LadeConfig)* - - type*: Literal['lade']* *= 'lade'* - - *class* qairt.modules.genie\_execution.genie\_config.LoraConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - adapters*: List[[LoraConfigAdapter](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LoraConfigAdapter)]* - - alpha\_tensor\_name*: Optional[str]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='alpha-tensor-name', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.LoraConfigAdapter(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - alphas*: List[str]* *= FieldInfo(annotation=NoneType, required=False, default\_factory=list)* - - bin\_sections*: List[str | os.PathLike]* *= FieldInfo(annotation=NoneType, required=False, default\_factory=list, alias='bin-sections', alias\_priority=2)* - - name*: str* - - *class* qairt.modules.genie\_execution.genie\_config.ModelBinary(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - ctx\_bins*: List[str | os.PathLike]* *= FieldInfo(annotation=NoneType, required=True, alias='ctx-bins', alias\_priority=2)* - - lora*: Optional[[LoraConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LoraConfig)]* *= None* - - *class* qairt.modules.genie\_execution.genie\_config.ModelLibrary(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - lora*: Optional[[LoraConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.LoraConfig)]* *= None* - - model\_bin*: str | os.PathLike* *= FieldInfo(annotation=NoneType, required=True, alias='model-bin', alias\_priority=2)* - - *pydantic model* qairt.modules.genie\_execution.genie\_config.PositionalEncoding - Bases: `BaseModel` - *field* rope\_dim*: Optional[int]* *= None* *(alias 'rope-dim')* - - *field* rope\_freqs\_type*: Optional[str]* *= None* *(alias 'rope-freqs-type')* - - *field* rope\_scaling*: Optional[[RopeScaling](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.RopeScaling)]* *= None* *(alias 'rope-scaling')* - - *field* rope\_theta*: Optional[float]* *= None* *(alias 'rope-theta')* - - *field* type*: Optional[[PositionalEncodingType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.PositionalEncodingType)]* *= None* - - *class* qairt.modules.genie\_execution.genie\_config.PositionalEncodingType(*value*) - Bases: `str`, `Enum` An enumeration. - ABSOLUTE *= 'absolute'* - - ALIBI *= 'alibi'* - - ROPE *= 'rope'* - - *class* qairt.modules.genie\_execution.genie\_config.QnnGenAiTransformerBackend(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - n\_embd*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-embd', alias\_priority=2)* - - n\_heads*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-heads', alias\_priority=2)* - - n\_layer*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-layer', alias\_priority=2)* - - n\_logits*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-logits', alias\_priority=2)* - - use\_mmap*: Optional[bool]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='use-mmap', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.QnnHtpBackend(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - allow\_async\_init*: Optional[bool]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='allow-async-init', alias\_priority=2)* - - cpu\_mask*: str* *= FieldInfo(annotation=NoneType, required=False, default='0x00', alias='cpu-mask', alias\_priority=2)* - - enable\_graph\_switching*: Optional[bool]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='enable-graph-switching', alias\_priority=2)* - - kv\_dim*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='kv-dim', alias\_priority=2)* - - kv\_update\_method*: Optional[str]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='kv-update-method', alias\_priority=2)* - - mmap\_budget*: int* *= FieldInfo(annotation=NoneType, required=False, default=40, alias='mmap-budget', alias\_priority=2)* - - poll*: bool* - - pos\_id\_dim*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='pos-id-dim', alias\_priority=2)* - - rope\_theta*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='rope-theta', alias\_priority=2)* - - spill\_fill\_bufsize*: int* *= FieldInfo(annotation=NoneType, required=False, default=0, alias='spill-fill-bufsize', alias\_priority=2)* - - use\_mmap*: bool* *= FieldInfo(annotation=NoneType, required=False, default=False, alias='use-mmap', alias\_priority=2)* - - *pydantic model* qairt.modules.genie\_execution.genie\_config.RopeScaling - Bases: `BaseModel` - *field* factor*: Optional[float]* *= None* - - *field* high\_freq\_factor*: Optional[float]* *= None* *(alias 'high-freq-factor')* - - *field* long\_factor*: Optional[List[float]]* *= None* *(alias 'long-factor')* - - *field* low\_freq\_factor*: Optional[float]* *= None* *(alias 'low-freq-factor')* - - *field* mrope\_section*: Optional[List[int]]* *= None* *(alias 'mrope-section')* - - *field* original\_max\_position\_embeddings*: Optional[int]* *= None* *(alias 'original-max-position-embeddings')* - - *field* rope\_type*: Optional[[RopeType](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.RopeType)]* *= None* *(alias 'rope-type')* - - *field* short\_factor*: Optional[List[float]]* *= None* *(alias 'short-factor')* - - *field* spatial\_merge\_size*: Optional[int]* *= None* *(alias 'spatial-merge-size')* - - *field* time\_step*: Optional[int]* *= None* *(alias 'time-step')* - - *class* qairt.modules.genie\_execution.genie\_config.RopeType(*value*) - Bases: `str`, `Enum` An enumeration. - DEFAULT *= 'default'* - - LLAMA3 *= 'llama3'* - - LONG\_ROPE *= 'longrope'* - - QWEN2VL *= 'qwen2vl'* - - QWEN2VL\_MROPE *= 'qwen2vl-mrope'* - - QWEN3VL *= 'qwen3vl'* - - QWEN3VL\_MROPE *= 'qwen3vl-mrope'* - - *class* qairt.modules.genie\_execution.genie\_config.SSDDialog(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`AbstractDialog`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.AbstractDialog) - engine*: [DialogEngine](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.DialogEngine)* - - ssd\_q1*: [SsdRunConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.SsdRunConfig)* *= FieldInfo(annotation=NoneType, required=True, alias='ssd-q1', alias\_priority=2)* - - type*: Literal['ssd-q1']* *= 'ssd-q1'* - - *class* qairt.modules.genie\_execution.genie\_config.Sampler(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - callback\_name*: Optional[str]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='callback-name', alias\_priority=2)* - - greedy*: Optional[bool]* *= None* - - seed*: int* *= 42* - - temp*: float* *= 0.8* - - top\_k*: int* *= FieldInfo(annotation=NoneType, required=False, default=40, alias='top-k', alias\_priority=2)* - - top\_p*: float* *= FieldInfo(annotation=NoneType, required=False, default=0.95, alias='top-p', alias\_priority=2)* - - type*: Optional[str]* *= None* - - *class* qairt.modules.genie\_execution.genie\_config.SsdConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - branches*: List[int]* - - forecast\_prefix*: int* *= FieldInfo(annotation=NoneType, required=True, alias='forecast-prefix', alias\_priority=2)* - - forecast\_token\_count*: int* *= FieldInfo(annotation=NoneType, required=True, alias='forecast-token-count', alias\_priority=2)* - - n\_streams*: Optional[int]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='n-streams', alias\_priority=2)* - - p\_threshold*: Optional[float]* *= FieldInfo(annotation=NoneType, required=False, default=None, alias='p-threshold', alias\_priority=2)* - - ssd\_version*: int* *= FieldInfo(annotation=NoneType, required=False, default=1, alias='ssd-version', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.SsdRunConfig(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`SsdConfig`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.SsdConfig) Configuration for running SSD inference. - forecast\_prefix\_name*: str | os.PathLike* *= FieldInfo(annotation=NoneType, required=True, alias='forecast-prefix-name', alias\_priority=2)* - - *class* qairt.modules.genie\_execution.genie\_config.Tokenizer(*\*args: Any*, *\*\*kwargs: Any*) - Bases: [`VersionedModel`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.VersionedModel) - path*: str | os.PathLike* *= ''* - - *class* qairt.modules.genie\_execution.genie\_config.VersionedModel(*\*args: Any*, *\*\*kwargs: Any*) - Bases: `AISWBaseModel` - version*: int* *= 1* - ## qairt.modules.genie\_execution.native\_t2t\_module - *class* qairt.modules.genie\_execution.native\_t2t\_module.GenieNativeT2TRunner(*genie\_config: [GenieConfig](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.genie_config.GenieConfig)*, *query\_timeout: int = 180*) - Bases: `StreamableGenieRunner` GenieNativeT2TRunner enables execution in native python environments using wrappers around Genie APIs. Users can utilize this API for access to more expressive and fine-grained control over execution in Genie. Implements `StreamableGenieRunner`, providing [`load()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.load), [`generate()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.generate), [`stream_generate()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.stream_generate), [`reset()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.reset), and [`unload()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.unload). - apply\_sampler\_config(*config: str*) → [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - Apply a sampler config either setting sampler parameters for the default sampler or supplying the name of a registered sampler callback function - Parameters - **config** (*str*) – json string representing the sampler config to apply - Returns - Returns self after applying the sampler config - Return type - self - extract\_profiling\_data() → Optional[GenerationMetrics] - - generate(*config: GenieQueryConfig*) → TextGenerationResult - generate(*config: GenieQueryConfig*, *\**, *generation\_config: Optional[GenerationConfig] = None*) → TextGenerationResult - generate(*config: str*) → TextGenerationResult - Executes the provided generation request. Accepts either a `GenieQueryConfig` (current API) or a plain prompt string (legacy API, deprecated). - Parameters - **config** (*Union* *[* *GenieQueryConfig* *,* *str* *]*) – Either a `GenieQueryConfig` containing the pre-formatted prompt and optional LoRA adapter configuration, **or** a plain prompt string (deprecated — use `GenieQueryConfig` instead). The native runner does not perform any chat template formatting, JSON parsing, or file loading — it passes the prompt directly to the underlying Genie dialog. Examples Current API: from qairt.modules.genie_execution.genie_config import GenieConfig from qairt.modules.genie_execution.native_t2t_module import GenieNativeT2TRunner with open("genie_config.json", "r") as f: genie_config = GenieConfig(**json.load(f)) runner = GenieNativeT2TRunner(genie_config) prompt = "<|begin_of_text|>What is the capital of Spain?:" result = runner.generate(GenieQueryConfig(prompt=prompt)) Copy to clipboard Legacy API (deprecated): result = runner.generate(prompt) Copy to clipboard - Returns - Generated text and execution metrics from native execution - Return type - TextGenerationResult - load() → None - No-op: native runner requires no device preparation. - register\_sampler\_callback(*name: str*, *callback: Callable[[ndarray], int]*) → [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - Register sampler callback function - Parameters - - **name** (*str*) – Name of the callback. Passed in a sampler config to set the desired sampler callback - **callback** (*Callable* *[* *[* *np.ndarray* *]* *,* *int* *]*) – Sampler callback to select the next token given logits - Returns - Returns self after registering callback - Return type - self - reset() → None - Reset accumulated dialog context so the next generation starts a fresh conversation. Delegates to [`reset_dialog()`](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner.reset_dialog). - reset\_dialog() → [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - Resets dialog state to remove context accumulated from queries. Also satisfies the `reset()` contract — callers may use either name. - Returns - Returns self after resetting dialog state. - Return type - [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - restore\_dialog(*saved\_dialog: Union[str, PathLike]*) → [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - Restores a saved genie dialog state - Parameters - **saved\_dialog** (*Union* *[* *str* *,* *os.PathLike* *]*) – Path to saved dialog state to restore - Returns - Returns self after restoration - Return type - [GenieNativeT2TRunner](https://docs.qualcomm.com/doc/80-87189-2/topic/qairt-gen-ai-modules-genie-execution.html#qairt.modules.genie_execution.native_t2t_module.GenieNativeT2TRunner) - save\_dialog(*save\_dir: Union[str, PathLike]*) → None - Stores the current state of the genie dialog - Parameters - **save\_dir** (*Union* *[* *str* *,* *os.PathLike* *]*) – Location to save the dialog - *async* stream\_generate(*prompt: str*, *streamer: Queue*, *\**, *generation\_config: Optional[GenerationConfig] = None*) → TextGenerationResult - Executes a prompt in streaming mode, sending output chunks to a queue as they are generated. This function runs the blocking LLM query in a background thread using run\_in\_executor, allowing the main event loop to remain responsive. As the model generates output, it invokes capture\_output to push each chunk into the provided asyncio queue. A None value is sent to the queue to signal the end of the stream. - Parameters - - **prompt** (*str*) – The input prompt to send to the LLM. - **streamer** (*asyncio.Queue*) – An asyncio queue used to stream output chunks back to the caller. - **generation\_config** – Optional per-call generation parameters that override the static defaults baked into the runner at load time. When `None` (the default) the runner uses its base configuration unchanged. - Returns - An object containing the full generated text and any execution metrics or errors. - Return type - TextGenerationResult - Raises - **asyncio.TimeoutError** – If the query exceeds the timeout duration. - unload() → None - No-op: resources are released by `__del__()`. ## qairt.modules.genie\_execution.genie\_app\_run\_module `GenieAppRunner` runs a Genie pipeline (including multimodal image-to-text pipelines) from a workflow and its compiled model containers. Use `GenieAppRunner.load()` to load the pipeline, `GenieAppRunner.generate()` to run inference, and `GenieAppRunner.unload()` to release resources. Configuration is passed via `GenieAppRunExecutionConfig` and output is described by `GenieAppRunOutputConfig`. Last Published: Aug 19, 2026 [Previous Topic Modules](https://docs.qualcomm.com/bundle/publicresource/80-87189-2/topics/qairt-modules.md) [Next Topic qairt.modules.lora](https://docs.qualcomm.com/bundle/publicresource/80-87189-2/topics/qairt-gen-ai-modules-lora.md)