# Image: winglian/axolotl:main-py3.10-cu118-2.0.1 
base_model: meta-llama/Llama-2-7b-hf
base_model_config: meta-llama/Llama-2-7b-hf
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: bentarnoff/logic_magazine_jsonl
    type: sharegpt
hub_model_id: bentarnoff/logic_magazine_jsonl
val_set_size: 0.01
output_dir: ./qlora-out

adapter: qlora
lora_model_dir:

sequence_len: 4096
sample_packing: false
pad_to_sequence_len: true

lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project: "logic_magazine"
wandb_entity:
wandb_watch:
wandb_run_id:
wandb_log_model: "checkpoint"

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 3
optimizer: paged_adamw_32bit
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

warmup_steps: 10
eval_steps: 20
eval_table_size: 5
save_steps:
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
  unk_token: "<unk>"

logic_magazine_jsonl

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the None dataset. It achieves the following results on the evaluation set:

Loss: 2.3642

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: 2
eval_batch_size: 2
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 4
total_train_batch_size: 32
total_eval_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 10
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
2.3553	0.15	20	2.4167
2.2767	0.31	40	2.3869
2.2854	0.46	60	2.3658
2.2849	0.61	80	2.3470
2.353	0.76	100	2.3337
2.2412	0.92	120	2.3363
2.1992	1.07	140	2.3240
2.1069	1.22	160	2.3404
2.2444	1.37	180	2.3403
2.1424	1.53	200	2.3446
2.1739	1.68	220	2.3404
2.1423	1.83	240	2.3382
2.1721	1.98	260	2.3378
2.1621	2.14	280	2.3630
2.0394	2.29	300	2.3623
2.0631	2.44	320	2.3665
2.0234	2.6	340	2.3632
2.1042	2.75	360	2.3654
2.02	2.9	380	2.3642

Framework versions

PEFT 0.7.0
Transformers 4.37.0.dev0
Pytorch 2.0.1+cu118
Datasets 2.16.1
Tokenizers 0.15.0

bentarnoff/logic_magazine_model

logic_magazine_jsonl

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

bentarnoff/
logic_magazine_model