KolaGang/finals

See axolotl config

axolotl version: 0.4.0

base_model: mistralai/Mistral-7B-Instruct-v0.2
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: YoungPanda/chatlaw
    type: sharegpt
  - path: KolaGang/legal_sum
    type: alpaca
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./pytorch

lisa_n_layers: 4
lisa_step_interval: 20
lisa_layers_attribute: model.layers

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

wandb_project: mistral_mistral
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 8
num_epochs: 3
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.000005

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

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint: out/checkpoint-99
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true
flash_attn_cross_entropy: false
flash_attn_rms_norm: true
flash_attn_fuse_qkv: false
flash_attn_fuse_mlp: true


warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
save_safetensors: False

pytorch

This model is a fine-tuned version of mistralai/Mistral-7B-Instruct-v0.2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7778

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: 5e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• total_eval_batch_size: 64
• 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
2.6768	0.03	1	4.0531
1.521	0.27	9	1.3495
1.1368	0.53	18	0.9795
1.0257	0.8	27	0.8902
0.9861	1.04	36	0.8528
0.9431	1.31	45	0.8288
0.94	1.58	54	0.8070
0.8841	1.84	63	0.7938
0.8442	2.09	72	0.7851
0.8251	2.36	81	0.7808
0.8591	2.62	90	0.7783
0.8369	2.89	99	0.7778

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.2.2
• Datasets 2.18.0
• Tokenizers 0.15.0