mm4.7-3b-slerp-comp-alpaca

<!-- 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/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>

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

axolotl version: 0.4.0

adapter: qlora
base_model: qlora-out2/merged
bf16: false
dataset_prepared_path: null
datasets:
- data_files:
  - Chain-of-Thought/formatted_cot_data/gsm8k_train.json
  path: QingyiSi/Alpaca-CoT
  type: alpaca:chat
debug: null
deepspeed: null
early_stopping_patience: null
evals_per_epoch: null
flash_attention: false
fp16: true
fsdp: null
fsdp_config: null
gradient_accumulation_steps: 1
gradient_checkpointing: true
group_by_length: false
is_llama_derived_model: true
learning_rate: 0.0003
load_in_4bit: true
load_in_8bit: false
local_rank: null
logging_steps: 1
lora_alpha: 16
lora_dropout: 0.05
lora_fan_in_fan_out: null
lora_model_dir: null
lora_r: 32
lora_target_linear: true
lora_target_modules: null
lr_scheduler: cosine
max_steps: 369
micro_batch_size: 1
mlflow_experiment_name: colab-example
model_type: LlamaForCausalLM
num_epochs: 6
optimizer: paged_adamw_32bit
output_dir: ./qlora-out
pad_to_sequence_len: true
resume_from_checkpoint: null
sample_packing: true
saves_per_epoch: null
sequence_len: 2048
special_tokens:
  bos_token: <s>
  eos_token: </s>
  unk_token: <unk>
strict: false
tf32: false
tokenizer_type: LlamaTokenizer
train_on_inputs: true
val_set_size: 0.2
wandb_entity: null
wandb_log_model: null
wandb_name: null
wandb_project: null
wandb_watch: null
warmup_steps: 10
weight_decay: 0.1
xformers_attention: false

</details><br>

qlora-out

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7597

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: 0.0003
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 369
• mixed_precision_training: Native AMP

Training results

Framework versions

• PEFT 0.8.2
• Transformers 4.39.0.dev0
• Pytorch 2.1.1+cu121
• Datasets 2.17.1
• Tokenizers 0.15.0