mixtral-pb-10e / README.md
dyang415's picture
End of training
2ce024f verified
metadata
license: apache-2.0
library_name: peft
tags:
  - axolotl
  - generated_from_trainer
base_model: mistralai/Mixtral-8x7B-Instruct-v0.1
model-index:
  - name: mixtral-pb-10e
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

base_model: mistralai/Mixtral-8x7B-Instruct-v0.1
model_type: AutoModelForCausalLM
tokenizer_type: LlamaTokenizer
trust_remote_code: true

load_in_8bit: false
load_in_4bit: true
strict: false
chat_template: inst

datasets:
  - path: ./data/pablo_processed.jsonl
    type: sharegpt
    conversation: mistral
  # - path: ./data/tool_used_training.jsonl
  #   type: sharegpt
  #   conversation: mistral
  # - path: ./data/tool_not_used_training.jsonl
  #   type: sharegpt
  #   conversation: mistral
  # - path: ./data/no_tools_training.jsonl
  #   type: sharegpt
  #   conversation: mistral

hub_model_id: dyang415/mixtral-pb-10e


dataset_prepared_path: last_run_prepared
val_set_size: 0.0
output_dir: ../mixtral-pb-10e

model_config:
  output_router_logits: true

adapter: qlora
lora_model_dir:

sequence_len: 16384
sample_packing: true
pad_to_sequence_len: true

lora_r: 8
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj


# wandb_project: function-call
# wandb_name: mixtral-instruct-lora--v1
# wandb_log_model: end
# hub_model_id: dyang415/mixtral-lora-v0


gradient_accumulation_steps: 2
micro_batch_size: 1
num_epochs: 10
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 0.0002

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

gradient_checkpointing: true
logging_steps: 1
flash_attention: true

loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
weight_decay: 0.0
fsdp:
fsdp_config:

mixtral-pb-10e

This model is a fine-tuned version of mistralai/Mixtral-8x7B-Instruct-v0.1 on the None dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

The following bitsandbytes quantization config was used during training:

  • quant_method: QuantizationMethod.BITS_AND_BYTES
  • load_in_8bit: False
  • load_in_4bit: True
  • llm_int8_threshold: 6.0
  • llm_int8_skip_modules: None
  • llm_int8_enable_fp32_cpu_offload: False
  • llm_int8_has_fp16_weight: False
  • bnb_4bit_quant_type: nf4
  • bnb_4bit_use_double_quant: True
  • bnb_4bit_compute_dtype: bfloat16

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: 2
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 4
  • total_eval_batch_size: 2
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 10

Training results

Framework versions

  • PEFT 0.7.0
  • Transformers 4.37.0
  • Pytorch 2.0.1+cu117
  • Datasets 2.17.1
  • Tokenizers 0.15.0