bridge.recipes.ministral3.h100.ministral3#

Ministral3 finetuning recipes with parameterless API.

This module provides SFT and PEFT configurations for Ministral3 models (3B, 8B, 14B).

Module Contents#

Functions#

ministral3_3b_sft_1gpu_h100_bf16_config

Return a full SFT config for Ministral3 3B.

ministral3_8b_sft_2gpu_h100_bf16_config

Return a full SFT config for Ministral3 8B.

ministral3_14b_sft_4gpu_h100_bf16_config

Return a full SFT config for Ministral3 14B.

ministral3_3b_peft_1gpu_h100_bf16_config

Return a PEFT config for Ministral3 3B.

ministral3_8b_peft_1gpu_h100_bf16_config

Return a PEFT config for Ministral3 8B.

ministral3_14b_peft_2gpu_h100_bf16_config

Return a PEFT config for Ministral3 14B.

Data#

API#

bridge.recipes.ministral3.h100.ministral3.ministral3_3b_sft_1gpu_h100_bf16_config() megatron.bridge.training.config.ConfigContainer#

Return a full SFT config for Ministral3 3B.

Default configuration: 1 GPU

  • TP=1, PP=1

  • LR=5e-6 (full SFT)

  • Sequence length: 4096

bridge.recipes.ministral3.h100.ministral3.ministral3_8b_sft_2gpu_h100_bf16_config() megatron.bridge.training.config.ConfigContainer#

Return a full SFT config for Ministral3 8B.

Default configuration: 2 GPUs

  • TP=2, PP=1

  • LR=5e-6 (full SFT)

  • Sequence length: 4096

bridge.recipes.ministral3.h100.ministral3.ministral3_14b_sft_4gpu_h100_bf16_config() megatron.bridge.training.config.ConfigContainer#

Return a full SFT config for Ministral3 14B.

Default configuration: 4 GPUs

  • TP=4, PP=1

  • LR=5e-6 (full SFT)

  • Sequence length: 4096

bridge.recipes.ministral3.h100.ministral3.ministral3_3b_peft_1gpu_h100_bf16_config(
peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',
) megatron.bridge.training.config.ConfigContainer#

Return a PEFT config for Ministral3 3B.

Default configuration: 1 GPU

  • TP=1, PP=1

  • LR=1e-4 (PEFT)

  • Sequence length: 4096

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

bridge.recipes.ministral3.h100.ministral3.ministral3_8b_peft_1gpu_h100_bf16_config(
peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',
) megatron.bridge.training.config.ConfigContainer#

Return a PEFT config for Ministral3 8B.

Default configuration: 1 GPU

  • TP=1, PP=1

  • LR=1e-4 (PEFT)

  • Sequence length: 4096

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

bridge.recipes.ministral3.h100.ministral3.ministral3_14b_peft_2gpu_h100_bf16_config(
peft_scheme: str | megatron.bridge.peft.base.PEFT = 'lora',
) megatron.bridge.training.config.ConfigContainer#

Return a PEFT config for Ministral3 14B.

Default configuration: 2 GPUs

  • TP=2, PP=1

  • LR=1e-4 (PEFT)

  • Sequence length: 4096

Parameters:

peft_scheme – PEFT scheme - “lora”, “dora”, or a custom PEFT instance.

bridge.recipes.ministral3.h100.ministral3.__all__#

[‘ministral3_14b_peft_2gpu_h100_bf16_config’, ‘ministral3_14b_sft_4gpu_h100_bf16_config’, ‘ministral…