base_model: Dans-DiscountModels/Mistral-Nemo-Base-2407-ChatML-Mod
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

trust_remote_code:

# wandb configuration
wandb_project: 12b-mn-dans-reasoning-test
wandb_watch:

wandb_run_id: V0.0.3-1-3 # V{Version}-{Run Number}-{Attempt Number}
wandb_log_model:

# push checkpoints to hub
hub_model_id: Dans-DiscountModels/12b-mn-dans-reasoning-test-4
# how to push checkpoints to hub
# https://huggingface.co/docs/transformers/v4.31.0/en/main_classes/trainer#transformers.TrainingArguments.hub_strategy
hub_strategy: "every_save"
# Whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets
# Required to be true when used in combination with `push_dataset_to_hub`
hf_use_auth_token: true

# where to save the finished model to
output_dir: ./12b-mn-dans-reasoning-test

save_safetensors: true

# dataset settings (local or huggingface repo)
datasets:
  - path: PocketDoc/Dans-Reasoningmaxx-NaturalReasoning
    type: dan-chat-advanced
  - path: PocketDoc/Dans-Reasoningmaxx-WebInstruct
    type: dan-chat-advanced
  - path: PocketDoc/Dans-Benchmaxx-COT
    type: dan-chat-advanced
  - path: PocketDoc/Dans-Logicmaxx-SAT-AP
    type: dan-chat-advanced
  - path: PocketDoc/Dans-Assistantmaxx-Opus-Merge
    type: dan-chat-advanced
  - path: PocketDoc/Dans-Assistantmaxx-sonnetorca-subset
    type: dan-chat-advanced

plugins:
  - axolotl.integrations.liger.LigerPlugin
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
liger_rope: true
liger_rms_norm: true
liger_layer_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: false
cut_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

adapter:
lora_model_dir:

lora_r: 128
lora_alpha: 128
lora_dropout: 0.1
lora_target_linear: True
lora_target_modules:
lora_modules_to_save:
  - embed_tokens
  - lm_head
lora_fan_in_fan_out:
peft_use_rslora: true

dataset_prepared_path: ./12b-mn-dans-reasoning-test-data
val_set_size: 0.005

sequence_len: 8192

sample_packing: true
eval_sample_packing: true

pad_to_sequence_len: true

gradient_checkpointing: true
# gradient_checkpointing_kwargs:
# use_reentrant: false

gradient_accumulation_steps: 1
micro_batch_size: 4

num_epochs: 2

optimizer: came_pytorch

lr_scheduler: rex
learning_rate: 0.0000015
cosine_min_lr_ratio: 0.1

weight_decay: 0.1

max_grad_norm: 0.1

train_on_inputs: false
group_by_length: true

bf16: true
fp16: false
tf32: false

early_stopping_patience:

resume_from_checkpoint:
auto_resume_from_checkpoints: true

local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_ratio: 0.05

evals_per_epoch: 16
eval_table_size:
eval_max_new_tokens:

saves_per_epoch: 8
save_total_limit: 1

debug: false

deepspeed: deepspeed_configs/zero3_bf16.json

fsdp:
fsdp_config:

special_tokens:

12b-mn-dans-reasoning-test-4

This model is a fine-tuned version of Dans-DiscountModels/Mistral-Nemo-Base-2407-ChatML-Mod on the PocketDoc/Dans-Reasoningmaxx-NaturalReasoning, the PocketDoc/Dans-Reasoningmaxx-WebInstruct, the PocketDoc/Dans-Benchmaxx-COT, the PocketDoc/Dans-Logicmaxx-SAT-AP, the PocketDoc/Dans-Assistantmaxx-Opus-Merge and the PocketDoc/Dans-Assistantmaxx-sonnetorca-subset datasets. It achieves the following results on the evaluation set:

Loss: 0.6204

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1.5e-06
train_batch_size: 4
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 4
total_train_batch_size: 16
total_eval_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_HF with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 38
num_epochs: 2.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.8049	0.0026	1	0.8338
0.8006	0.0644	25	0.7551
0.7197	0.1289	50	0.7009
0.7165	0.1933	75	0.6810
0.7183	0.2577	100	0.6697
0.6671	0.3222	125	0.6620
0.6406	0.3866	150	0.6567
0.6656	0.4510	175	0.6528
0.6539	0.5155	200	0.6489
0.634	0.5799	225	0.6460
0.6606	0.6443	250	0.6428
0.6815	0.7088	275	0.6401
0.6082	0.7732	300	0.6385
0.6754	0.8376	325	0.6364
0.6284	0.9021	350	0.6347
0.6517	0.9665	375	0.6326
0.5583	1.0309	400	0.6340
0.5716	1.0954	425	0.6328
0.5799	1.1598	450	0.6323
0.5957	1.2242	475	0.6316
0.589	1.2887	500	0.6300
0.6007	1.3531	525	0.6289
0.5751	1.4175	550	0.6284
0.5627	1.4820	575	0.6275
0.5689	1.5464	600	0.6267
0.5098	1.6108	625	0.6259
0.5623	1.6753	650	0.6248
0.5608	1.7397	675	0.6237
0.5552	1.8041	700	0.6233
0.554	1.8686	725	0.6228
0.5696	1.9330	750	0.6217
0.5537	1.9974	775	0.6204

Framework versions

Transformers 4.49.0
Pytorch 2.4.1+cu124
Datasets 3.2.0
Tokenizers 0.21.1