Upload 2 files
Browse files
Erebus_Gemma_Control_Completion.yaml
ADDED
|
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
base_model: unsloth/gemma-2-9b
|
| 2 |
+
model_type: AutoModelForCausalLM
|
| 3 |
+
tokenizer_type: AutoTokenizer
|
| 4 |
+
|
| 5 |
+
hub_model_id: NewEden/Erebus-Control-9B
|
| 6 |
+
hub_strategy: "all_checkpoints"
|
| 7 |
+
push_dataset_to_hub:
|
| 8 |
+
hf_use_auth_token: true
|
| 9 |
+
|
| 10 |
+
plugins:
|
| 11 |
+
- axolotl.integrations.liger.LigerPlugin
|
| 12 |
+
liger_rope: true
|
| 13 |
+
liger_rms_norm: true
|
| 14 |
+
liger_swiglu: true
|
| 15 |
+
#liger_cross_entropy: true
|
| 16 |
+
liger_fused_linear_cross_entropy: true
|
| 17 |
+
|
| 18 |
+
load_in_8bit: false
|
| 19 |
+
load_in_4bit: false
|
| 20 |
+
strict: false
|
| 21 |
+
|
| 22 |
+
datasets:
|
| 23 |
+
- path: Mielikki/Erebus-87k
|
| 24 |
+
type: completion
|
| 25 |
+
field: body
|
| 26 |
+
# - path: NewEden/4chan-smol-sharegpt
|
| 27 |
+
# type: sharegpt
|
| 28 |
+
# conversation: chatml
|
| 29 |
+
# - path: anthracite-org/kalo-opus-instruct-22k-no-refusal
|
| 30 |
+
# type: sharegpt
|
| 31 |
+
# conversation: chatml
|
| 32 |
+
# - path: NewEden/Claude-Instruct-2.7k
|
| 33 |
+
# type: sharegpt
|
| 34 |
+
# conversation: chatml
|
| 35 |
+
# - path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
|
| 36 |
+
# type: sharegpt
|
| 37 |
+
# conversation: chatml
|
| 38 |
+
# - path: NewEden/Claude-Instruct-5K
|
| 39 |
+
# type: sharegpt
|
| 40 |
+
# conversation: chatml
|
| 41 |
+
#chat_template: chatml
|
| 42 |
+
shuffle_merged_datasets: true
|
| 43 |
+
#default_system_message: "You are an assistant that responds to the user."
|
| 44 |
+
dataset_prepared_path: Control-9B
|
| 45 |
+
val_set_size: 0.0
|
| 46 |
+
output_dir: Control-9B
|
| 47 |
+
|
| 48 |
+
sequence_len: 32768
|
| 49 |
+
sample_packing: true
|
| 50 |
+
pad_to_sequence_len: true
|
| 51 |
+
|
| 52 |
+
wandb_project: 14b-erebus
|
| 53 |
+
wandb_entity:
|
| 54 |
+
wandb_watch:
|
| 55 |
+
wandb_name: r3-base-attempt-01
|
| 56 |
+
wandb_log_model:
|
| 57 |
+
|
| 58 |
+
gradient_accumulation_steps: 4
|
| 59 |
+
micro_batch_size: 1
|
| 60 |
+
num_epochs: 2
|
| 61 |
+
optimizer: adamw_bnb_8bit
|
| 62 |
+
lr_scheduler: cosine
|
| 63 |
+
learning_rate: 0.00001
|
| 64 |
+
|
| 65 |
+
train_on_inputs: false
|
| 66 |
+
group_by_length: false
|
| 67 |
+
bf16: auto
|
| 68 |
+
fp16:
|
| 69 |
+
tf32: false
|
| 70 |
+
|
| 71 |
+
gradient_checkpointing: true
|
| 72 |
+
early_stopping_patience:
|
| 73 |
+
resume_from_checkpoint:
|
| 74 |
+
local_rank:
|
| 75 |
+
logging_steps: 1
|
| 76 |
+
xformers_attention:
|
| 77 |
+
flash_attention: true
|
| 78 |
+
|
| 79 |
+
warmup_steps: 5
|
| 80 |
+
evals_per_epoch:
|
| 81 |
+
eval_table_size:
|
| 82 |
+
eval_max_new_tokens:
|
| 83 |
+
saves_per_epoch: 2
|
| 84 |
+
debug:
|
| 85 |
+
deepspeed: deepspeed_configs/zero2.json
|
| 86 |
+
weight_decay: 0.01
|
| 87 |
+
fsdp:
|
| 88 |
+
fsdp_config:
|
| 89 |
+
special_tokens:
|
| 90 |
+
pad_token: <|endoftext|>
|
| 91 |
+
eos_token: <|endoftext|>
|
Erebus_Gemma_control_Inst.yaml
ADDED
|
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
base_model: NewEden/Erebus-Control-9B
|
| 2 |
+
model_type: AutoModelForCausalLM
|
| 3 |
+
tokenizer_type: AutoTokenizer
|
| 4 |
+
|
| 5 |
+
hub_model_id: NewEden/Erebus-Control-9B-Final
|
| 6 |
+
hub_strategy: "all_checkpoints"
|
| 7 |
+
push_dataset_to_hub:
|
| 8 |
+
hf_use_auth_token: true
|
| 9 |
+
|
| 10 |
+
plugins:
|
| 11 |
+
- axolotl.integrations.liger.LigerPlugin
|
| 12 |
+
liger_rope: true
|
| 13 |
+
liger_rms_norm: true
|
| 14 |
+
liger_swiglu: true
|
| 15 |
+
#liger_cross_entropy: true
|
| 16 |
+
liger_fused_linear_cross_entropy: true
|
| 17 |
+
|
| 18 |
+
load_in_8bit: false
|
| 19 |
+
load_in_4bit: false
|
| 20 |
+
strict: false
|
| 21 |
+
|
| 22 |
+
datasets:
|
| 23 |
+
- path: NewEden/4chan-smol-sharegpt
|
| 24 |
+
type: sharegpt
|
| 25 |
+
conversation: chatml
|
| 26 |
+
- path: anthracite-org/kalo-opus-instruct-22k-no-refusal
|
| 27 |
+
type: sharegpt
|
| 28 |
+
conversation: chatml
|
| 29 |
+
- path: NewEden/Claude-Instruct-2.7k
|
| 30 |
+
type: sharegpt
|
| 31 |
+
conversation: chatml
|
| 32 |
+
- path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
|
| 33 |
+
type: sharegpt
|
| 34 |
+
conversation: chatml
|
| 35 |
+
- path: NewEden/Claude-Instruct-5K
|
| 36 |
+
type: sharegpt
|
| 37 |
+
conversation: chatml
|
| 38 |
+
- path: NewEden/OpenCAI-ShareGPT
|
| 39 |
+
type: sharegpt
|
| 40 |
+
conversation: chatml
|
| 41 |
+
- path: NewEden/Roleplay-Logs-Sharegpt-Ngram-cleaned
|
| 42 |
+
type: sharegpt
|
| 43 |
+
conversation: chatml
|
| 44 |
+
- path: NewEden/PIPPA-Mega-Filtered
|
| 45 |
+
type: sharegpt
|
| 46 |
+
conversation: chatml
|
| 47 |
+
#chat_template: chatml
|
| 48 |
+
shuffle_merged_datasets: true
|
| 49 |
+
#default_system_message: "You are an assistant that responds to the user."
|
| 50 |
+
dataset_prepared_path: Control-9B
|
| 51 |
+
val_set_size: 0.0
|
| 52 |
+
output_dir: Control-9B
|
| 53 |
+
|
| 54 |
+
sequence_len: 8192
|
| 55 |
+
sample_packing: true
|
| 56 |
+
pad_to_sequence_len: true
|
| 57 |
+
|
| 58 |
+
wandb_project: Erebus-Control-9b
|
| 59 |
+
wandb_entity:
|
| 60 |
+
wandb_watch:
|
| 61 |
+
wandb_name: attempt-1
|
| 62 |
+
wandb_log_model:
|
| 63 |
+
|
| 64 |
+
gradient_accumulation_steps: 4
|
| 65 |
+
micro_batch_size: 1
|
| 66 |
+
num_epochs: 4
|
| 67 |
+
optimizer: adamw_bnb_8bit
|
| 68 |
+
lr_scheduler: cosine
|
| 69 |
+
learning_rate: 0.00001
|
| 70 |
+
|
| 71 |
+
train_on_inputs: false
|
| 72 |
+
group_by_length: false
|
| 73 |
+
bf16: auto
|
| 74 |
+
fp16:
|
| 75 |
+
tf32: false
|
| 76 |
+
|
| 77 |
+
gradient_checkpointing: true
|
| 78 |
+
early_stopping_patience:
|
| 79 |
+
resume_from_checkpoint:
|
| 80 |
+
local_rank:
|
| 81 |
+
logging_steps: 1
|
| 82 |
+
xformers_attention:
|
| 83 |
+
flash_attention: true
|
| 84 |
+
|
| 85 |
+
warmup_steps: 10
|
| 86 |
+
evals_per_epoch:
|
| 87 |
+
eval_table_size:
|
| 88 |
+
eval_max_new_tokens:
|
| 89 |
+
saves_per_epoch: 2
|
| 90 |
+
debug:
|
| 91 |
+
deepspeed: deepspeed_configs/zero2.json
|
| 92 |
+
weight_decay: 0.001
|
| 93 |
+
fsdp:
|
| 94 |
+
fsdp_config:
|
| 95 |
+
special_tokens:
|
| 96 |
+
pad_token: <pad>
|