reasoning-llama-3.1-70b-stratos-cold-start-v2

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>

<details><summary>See axolotl config</summary>

axolotl version: 0.8.0.dev0

base_model: meta-llama/Llama-3.1-70B
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name
#

plugins:
  - axolotl.integrations.liger.LigerPlugin
  - axolotl.integrations.spectrum.SpectrumPlugin
spectrum_top_fraction: 0.5
spectrum_model_name: meta-llama/Meta-Llama-3.1-70B
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
liger_fused_linear_cross_entropy: true

strict: false

chat_template: llama3
datasets:
  - path: bespokelabs/Bespoke-Stratos-17k
    field_messages: conversations
    message_property_mappings:
      content: value
      role: from
    split: train
    type: chat_template
dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ./outputs/out/reasoning-70b-stratos
save_safetensors: true

wandb_project: reasoning-70b-stratos
wandb_entity: axolotl-ai
wandb_watch:
wandb_name:
wandb_log_model:

sequence_len: 16384
sample_packing: true
pad_to_sequence_len: true

gradient_accumulation_steps: 1
micro_batch_size: 4
num_epochs: 3
optimizer: adamw_torch_fused
lr_scheduler: rex
learning_rate: 5.0e-6
embedding_lr: 1.0e-5
max_grad_norm: 1.0

train_on_inputs: false
group_by_length: false
bf16: true
tf32: true

gradient_checkpointing: offload
gradient_checkpointing_kwargs:
  use_reentrant: true
logging_steps: 1
flash_attention: true

warmup_steps: 20
evals_per_epoch: 4
saves_per_epoch: 1
weight_decay: 0.01
deepspeed: deepspeed_configs/zero3_bf16_cpuoffload_params.json
special_tokens:
  pad_token: <|finetune_right_pad_id|>
  eos_token: <|eot_id|>
added_tokens_overrides:
  128011: <think>
  128012: </think>
  128013: <|begin_of_thought|>
  128014: <|end_of_thought|>
  128015: <|begin_of_solution|>
  128016: <|end_of_solution|>
fix_untrained_tokens:
  - 128011
  - 128012
  - 128013
  - 128014
  - 128015
  - 128016

</details><br>

outputs/out/reasoning-70b-stratos

This model is a fine-tuned version of meta-llama/Llama-3.1-70B on the bespokelabs/Bespoke-Stratos-17k dataset.

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: 5e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 16
• total_train_batch_size: 64
• total_eval_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: rex
• lr_scheduler_warmup_steps: 20
• num_epochs: 3.0

Training results

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0