Francois-PE-12B

Name: Francois-PE-12B
Author: Delta-Vector

12.0B

—

Delta-Vector

Other

OTHER

12B params

New

0 downloads

Early-stage

Try on Hugging Face Add to Compare

Edge AI:

Mobile

Laptop

Server

27GB+ RAM

Mobile

Laptop

Server

Quick Summary

A finetune of Pocketdoc's Personality Engine with a data mix similar to Ohashi-NeMo.

Device Compatibility

Mobile

4-6GB RAM

Laptop

16GB RAM

Server

GPU

Minimum Recommended

12GB+ RAM

Code Examples

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Axolotl configyaml

base_model: PocketDoc_Dans-PersonalityEngine-V1.1.0-12b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: NewEden/OpenCAI-ShareGPT
    type: sharegpt
  - path: NewEden/vanilla-backrooms-claude-sharegpt
    type: sharegpt
  - path: anthracite-org/kalo_opus_misc_240827
    type: sharegpt
  - path: anthracite-org/kalo_misc_part2
    type: sharegpt
  - path: NewEden/RP-logs-V2-Experimental
    type: sharegpt
  - path: NewEden/Misc-Mang-Sharegpt
    type: sharegpt
  - path: NewEden/BlueSky-Experimental-sharegpt
    type: sharegpt

dataset_prepared_path: dataset_prepared
val_set_size: 0.0
output_dir: 12b-out-r2

sequence_len: 24576
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:
lora_r: 128
lora_alpha: 16
lora_dropout: 0.05 
peft_use_rslora: true
lora_target_modules:
  - gate_proj
  - down_proj
  - up_proj
  - q_proj
  - v_proj
  - k_proj
  - o_proj

lora_modules_to_save:
 - embed_tokens
 - lm_head


wandb_project: 12b-control
wandb_entity:
wandb_watch:
wandb_name: 12b-control-r2
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 1
num_epochs: 4
optimizer: paged_ademamix_8bit
# optimizer: paged_adamw_8bit

lr_scheduler: cosine
learning_rate: 0.00001

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: unsloth
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 40
evals_per_epoch: 
eval_table_size:
eval_max_new_tokens:
saves_per_epoch: 1
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16_cpuoffload_params.json
weight_decay: 0.03
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

Deploy This Model

Production-ready deployment in minutes

Together.ai

Instant API access to this model

Fastest API

Production-ready inference API. Start free, scale to millions.

Try Free API

Replicate

One-click model deployment

Easiest Setup

Run models in the cloud with simple API. No DevOps required.

Deploy Now

Disclosure: We may earn a commission from these partners. This helps keep LLMYourWay free.