tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

simplify the example configs to be more minimal and less daunting (#2486) [skip ci]

Browse Source 
* simplify the example configs to be more minimal and less daunting

* drop empty s2_attention from example yamls
This commit is contained in:
Wing Lian
2025-04-04 13:47:26 -04:00
committed by GitHub
parent dd66fb163c
commit 9f824ef76a
101 changed files with 14 additions and 1140 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
examples/llama-2/fft_optimized.yml
View File

@@ -4,9 +4,6 @@ model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
@@ -26,7 +23,6 @@ lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
@@ -41,18 +37,12 @@ optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
flash_attn_cross_entropy: false
flash_attn_rms_norm: true
@@ -61,11 +51,8 @@ flash_attn_fuse_mlp: true
warmup_steps: 100
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed: #deepspeed_configs/zero2.json # multi-gpu only
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:

10
examples/llama-2/gptq-lora.yml
View File

@@ -10,8 +10,6 @@ gptq_disable_exllama: true
tokenizer_use_fast: true
tokenizer_legacy: true
load_in_8bit: false
load_in_4bit: false
strict: false
push_dataset_to_hub:
hf_use_auth_token: true
@@ -33,7 +31,6 @@ lora_target_modules:
- q_proj
- v_proj
lora_target_linear:
lora_fan_in_fan_out:
wandb_project:
wandb_watch:
wandb_name:
@@ -50,26 +47,19 @@ torchdistx_path:
lr_scheduler: cosine
lr_quadratic_warmup: true
learning_rate: 0.000017
train_on_inputs: false
group_by_length: false
bf16: false
fp16: false
float16: true
tf32: true
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention:
sdp_attention:
flash_optimum:
warmup_steps: 100
evals_per_epoch: 4
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
special_tokens:
bos_token: "<s>"

15
examples/llama-2/lisa.yml
View File

@@ -4,9 +4,6 @@ model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
@@ -26,7 +23,6 @@ lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:
lisa_n_layers: 4
lisa_step_interval: 20
@@ -45,18 +41,12 @@ optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 5e-5 # recommendation from lisa paper for 7b
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
flash_attn_cross_entropy: false
flash_attn_rms_norm: true
@@ -65,13 +55,8 @@ flash_attn_fuse_mlp: true
warmup_steps: 100
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
bos_token: "<s>"
eos_token: "</s>"

17
examples/llama-2/loftq.yml
View File

@@ -4,9 +4,6 @@ model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name
load_in_8bit: false
load_in_4bit: false
strict: false
datasets:
@@ -26,7 +23,6 @@ lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
peft:
loftq_config:
loftq_bits: 4
@@ -44,29 +40,16 @@ optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

14
examples/llama-2/lora.yml
View File

@@ -26,7 +26,6 @@ lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
@@ -41,29 +40,16 @@ optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
s2_attention:
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

11
examples/llama-2/qlora-fsdp.yml
View File

@@ -26,9 +26,7 @@ pad_to_sequence_len: true
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
@@ -43,28 +41,19 @@ optimizer: adamw_torch_fused
lr_scheduler: cosine
learning_rate: 0.00001
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
- full_shard

13
examples/llama-2/qlora.yml
View File

@@ -26,9 +26,7 @@ pad_to_sequence_len: true
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
@@ -43,27 +41,16 @@ optimizer: paged_adamw_32bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:

12
examples/llama-2/relora.yml
View File

@@ -24,9 +24,7 @@ pad_to_sequence_len: true
lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:
relora_steps: 150
relora_warmup_steps: 10
@@ -45,28 +43,18 @@ optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 4
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
bos_token: "<s>"
eos_token: "</s>"