AQIT 0.1.0
Loading...
Searching...
No Matches
aquin.compute.train_simulate.SimulateRequest Class Reference
Inheritance diagram for aquin.compute.train_simulate.SimulateRequest:
Collaboration diagram for aquin.compute.train_simulate.SimulateRequest:

Static Public Attributes

str model_id = "meta-llama/Llama-3.2-1B-Instruct"
int rank = 8
int alpha = 16
float lr = 2e-4
int epochs = 3
float dropout = 0.05
list target_modules = []
int warmup_steps = 0
float grad_clip = 1.0
float weight_decay = 0.01
int grad_accum_steps = 1
str optimizer = "adamw"
str scheduler = "cosine"
int max_seq_len = 128
list dataset = None
bool use_qlora = False
bool use_full_ft = False
bool use_sft = False
bool use_dpo = False
float dpo_beta = 0.1
bool use_cpt = False
bool use_distil = False
str teacher_model_id = ""
float distil_temperature = 4.0
float distil_alpha = 0.7
bool use_rlhf = False
list preference_dataset = []
float rlhf_beta = 0.1

Detailed Description

Definition at line 79 of file train_simulate.py.

Member Data Documentation

◆ alpha

int aquin.compute.train_simulate.SimulateRequest.alpha = 16
static

Definition at line 82 of file train_simulate.py.

◆ dataset

list aquin.compute.train_simulate.SimulateRequest.dataset = None
static

Definition at line 94 of file train_simulate.py.

◆ distil_alpha

float aquin.compute.train_simulate.SimulateRequest.distil_alpha = 0.7
static

Definition at line 104 of file train_simulate.py.

◆ distil_temperature

float aquin.compute.train_simulate.SimulateRequest.distil_temperature = 4.0
static

Definition at line 103 of file train_simulate.py.

◆ dpo_beta

float aquin.compute.train_simulate.SimulateRequest.dpo_beta = 0.1
static

Definition at line 99 of file train_simulate.py.

◆ dropout

float aquin.compute.train_simulate.SimulateRequest.dropout = 0.05
static

Definition at line 85 of file train_simulate.py.

◆ epochs

int aquin.compute.train_simulate.SimulateRequest.epochs = 3
static

Definition at line 84 of file train_simulate.py.

◆ grad_accum_steps

int aquin.compute.train_simulate.SimulateRequest.grad_accum_steps = 1
static

Definition at line 90 of file train_simulate.py.

◆ grad_clip

float aquin.compute.train_simulate.SimulateRequest.grad_clip = 1.0
static

Definition at line 88 of file train_simulate.py.

◆ lr

float aquin.compute.train_simulate.SimulateRequest.lr = 2e-4
static

Definition at line 83 of file train_simulate.py.

◆ max_seq_len

int aquin.compute.train_simulate.SimulateRequest.max_seq_len = 128
static

Definition at line 93 of file train_simulate.py.

◆ model_id

str aquin.compute.train_simulate.SimulateRequest.model_id = "meta-llama/Llama-3.2-1B-Instruct"
static

Definition at line 80 of file train_simulate.py.

◆ optimizer

str aquin.compute.train_simulate.SimulateRequest.optimizer = "adamw"
static

Definition at line 91 of file train_simulate.py.

◆ preference_dataset

list aquin.compute.train_simulate.SimulateRequest.preference_dataset = []
static

Definition at line 107 of file train_simulate.py.

◆ rank

int aquin.compute.train_simulate.SimulateRequest.rank = 8
static

Definition at line 81 of file train_simulate.py.

◆ rlhf_beta

float aquin.compute.train_simulate.SimulateRequest.rlhf_beta = 0.1
static

Definition at line 108 of file train_simulate.py.

◆ scheduler

str aquin.compute.train_simulate.SimulateRequest.scheduler = "cosine"
static

Definition at line 92 of file train_simulate.py.

◆ target_modules

list aquin.compute.train_simulate.SimulateRequest.target_modules = []
static

Definition at line 86 of file train_simulate.py.

◆ teacher_model_id

str aquin.compute.train_simulate.SimulateRequest.teacher_model_id = ""
static

Definition at line 102 of file train_simulate.py.

◆ use_cpt

bool aquin.compute.train_simulate.SimulateRequest.use_cpt = False
static

Definition at line 100 of file train_simulate.py.

◆ use_distil

bool aquin.compute.train_simulate.SimulateRequest.use_distil = False
static

Definition at line 101 of file train_simulate.py.

◆ use_dpo

bool aquin.compute.train_simulate.SimulateRequest.use_dpo = False
static

Definition at line 98 of file train_simulate.py.

◆ use_full_ft

bool aquin.compute.train_simulate.SimulateRequest.use_full_ft = False
static

Definition at line 96 of file train_simulate.py.

◆ use_qlora

bool aquin.compute.train_simulate.SimulateRequest.use_qlora = False
static

Definition at line 95 of file train_simulate.py.

◆ use_rlhf

bool aquin.compute.train_simulate.SimulateRequest.use_rlhf = False
static

Definition at line 106 of file train_simulate.py.

◆ use_sft

bool aquin.compute.train_simulate.SimulateRequest.use_sft = False
static

Definition at line 97 of file train_simulate.py.

◆ warmup_steps

int aquin.compute.train_simulate.SimulateRequest.warmup_steps = 0
static

Definition at line 87 of file train_simulate.py.

◆ weight_decay

float aquin.compute.train_simulate.SimulateRequest.weight_decay = 0.01
static

Definition at line 89 of file train_simulate.py.


The documentation for this class was generated from the following file: