bridge.recipes.nemotronh.nemotron_3_ultra#
Module Contents#
Functions#
Return a pre-training config for Nemotron 3 Ultra. |
|
Return a packed OpenMathInstruct-2 full SFT config for Nemotron 3 Ultra. |
|
Return a packed OpenMathInstruct-2 PEFT config for Nemotron 3 Ultra. |
Data#
API#
- bridge.recipes.nemotronh.nemotron_3_ultra.NEMOTRON_3_ULTRA_HF_MODEL_ID#
ānvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16ā
- bridge.recipes.nemotronh.nemotron_3_ultra.NEMOTRON_3_ULTRA_TOKENIZER_NAME#
ānvidiaāNVIDIA-Nemotron-3-Ultra-550B-A55B-BF16ā
- bridge.recipes.nemotronh.nemotron_3_ultra.nemotron_3_ultra_pretrain_config(
- *,
- hf_path: str | None = None,
- seq_length: int = 8192,
Return a pre-training config for Nemotron 3 Ultra.
- Parameters:
hf_path ā Optional Hugging Face model ID or local snapshot path.
seq_length ā Sequence length for model and dataset settings.
- Returns:
Pre-training configuration for Nemotron 3 Ultra.
- bridge.recipes.nemotronh.nemotron_3_ultra.nemotron_3_ultra_sft_openmathinstruct2_packed_config(
- *,
- hf_path: str | None = None,
- seq_length: int = 4096,
Return a packed OpenMathInstruct-2 full SFT config for Nemotron 3 Ultra.
- Parameters:
hf_path ā Optional Hugging Face model ID or local snapshot path.
seq_length ā Packed sequence length.
- Returns:
Full-parameter SFT configuration for OpenMathInstruct-2.
- bridge.recipes.nemotronh.nemotron_3_ultra.nemotron_3_ultra_peft_openmathinstruct2_packed_config(
- *,
- peft: str | megatron.bridge.peft.base.PEFT | None = 'lora',
- hf_path: str | None = None,
- seq_length: int = 4096,
Return a packed OpenMathInstruct-2 PEFT config for Nemotron 3 Ultra.
- Parameters:
peft ā PEFT scheme, PEFT instance, or ānoneā.
hf_path ā Optional Hugging Face model ID or local snapshot path.
seq_length ā Packed sequence length.
- Returns:
PEFT configuration for OpenMathInstruct-2.
- bridge.recipes.nemotronh.nemotron_3_ultra.__all__#
[ānemotron_3_ultra_pretrain_configā, ānemotron_3_ultra_sft_openmathinstruct2_packed_configā, ānemotrā¦