Qwen2-37B-Pruned

<!-- 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: 0.4.1

base_model: carsonpoole/Qwen2-72B-Instruct-Every-Other-Layer
trust_remote_code: false

load_in_8bit: false
load_in_4bit: true # Changed to false
strict: false

datasets:
  - path: teknium/OpenHermes-2.5
    # type:
    #   system_prompt: ""
    #   system_format: "{system}"
    #   field_system: system
    #   field_instruction: human
    #   field_input: system
    #   field_output: gpt
    #   format: |-
    #     User: {instruction} {input}
    #     Assistant:
    #   no_input_format: "{instruction} "
    type: sharegpt

dataset_prepared_path:
val_set_size: 0
output_dir: ./outputs/out

sequence_len: 8192
sample_packing: true
eval_sample_packing: true
pad_to_sequence_len: true

hub_model_id: carsonpoole/Qwen2-37B-Pruned # Specify the repository name for pushing to the Hub
hub_strategy: every_save # Push checkpoint to the Hub on every save

wandb_project: forefront-research
wandb_entity: helloforefront
wandb_watch:
wandb_name: qwen2-37b-pruned-v0_3_2
wandb_log_model:

adapter: qlora
lora_model_dir:
lora_r: 32
lora_alpha: 64
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

gradient_accumulation_steps: 4
micro_batch_size: 1
num_epochs: 1
optimizer: adamw_torch
lr_scheduler: cosine
learning_rate: 0.0002

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

hf_use_auth_token: true

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 10
max_steps: 192
debug:
deepspeed:
weight_decay: 0.0
fsdp:
  - full_shard
  - auto_wrap
fsdp_config:
  fsdp_limit_all_gathers: true
  fsdp_sync_module_states: true
  fsdp_offload_params: true
  fsdp_use_orig_params: false
  fsdp_cpu_ram_efficient_loading: true
  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
  fsdp_transformer_layer_cls_to_wrap: Qwen2DecoderLayer
  fsdp_state_dict_type: FULL_STATE_DICT
special_tokens:

</details><br>

Qwen2-37B-Pruned

This model is a fine-tuned version of carsonpoole/Qwen2-72B-Instruct-Every-Other-Layer on the None 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: 0.0002
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 10
• training_steps: 192

Training results

Framework versions

• PEFT 0.11.1
• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1