axolotl version: 0.3.0
base_model: mistralai/Mistral-7B-v0.1
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer
is_mistral_derived_model: true
load_in_8bit: false
load_in_4bit: true
strict: false
datasets:
- path: theory_of_mind_airoboros_fixed.json
type: alpaca
dataset_prepared_path: last_run_prepared
val_set_size: 0.1
output_dir: ./qlora-out
adapter: qlora
lora_model_dir:
sequence_len: 2048
sample_packing: false
pad_to_sequence_len: true
lora_r: 128
lora_alpha: 256
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:
lora_target_modules:
- gate_proj
- down_proj
- up_proj
- q_proj
- v_proj
- k_proj
- o_proj
wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
gradient_accumulation_steps: 4
micro_batch_size: 4
num_epochs: 5
optimizer: adamw_bnb_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
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
loss_watchdog_threshold: 5.0
loss_watchdog_patience: 3
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_table_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
bos_token: "<s>"
eos_token: "</s>"
unk_token: "<unk>"
qlora-out
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.0709
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: bitsandbytes
- 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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.3182 | 0.05 | 1 | 1.3547 |
| 1.2887 | 0.26 | 5 | 1.2017 |
| 1.1629 | 0.52 | 10 | 1.2319 |
| 1.1734 | 0.78 | 15 | 1.2364 |
| 0.6007 | 1.04 | 20 | 1.3146 |
| 0.4225 | 1.3 | 25 | 1.4244 |
| 0.4144 | 1.56 | 30 | 1.4335 |
| 0.5067 | 1.82 | 35 | 1.4505 |
| 0.225 | 2.08 | 40 | 1.4928 |
| 0.1646 | 2.34 | 45 | 1.7377 |
| 0.1838 | 2.6 | 50 | 1.6058 |
| 0.2294 | 2.86 | 55 | 1.6419 |
| 0.0626 | 3.12 | 60 | 1.8140 |
| 0.0383 | 3.38 | 65 | 2.0478 |
| 0.0529 | 3.64 | 70 | 1.9511 |
| 0.0555 | 3.9 | 75 | 1.9203 |
| 0.0112 | 4.16 | 80 | 1.9597 |
| 0.0152 | 4.42 | 85 | 2.0307 |
| 0.0154 | 4.68 | 90 | 2.0652 |
| 0.0147 | 4.94 | 95 | 2.0709 |
Framework versions
- PEFT 0.7.0
- Transformers 4.37.0.dev0
- Pytorch 2.0.1+cu117
- Datasets 2.16.1
- Tokenizers 0.15.0