base_model: chargoddard/internlm2-20b-llama
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true

load_in_8bit: true
load_in_4bit: false
strict: false

datasets:
  - path: ARB/arb_law.json
    ds_type: json
    type: alpaca
    conversation: chatml

  - path: ARB/arb_math.json
    ds_type: json
    type: alpaca
    conversation: chatml

  - path: ARB/arb_mcat_reading.json
    ds_type: json
    type: alpaca
    conversation: chatml

  - path: ARB/arb_mcat_science.json
    ds_type: json
    type: alpaca
    conversation: chatml

  - path: ARB/arb_physics.json
    ds_type: json
    type: alpaca
    conversation: chatml


dataset_prepared_path: last_run_prepared
val_set_size: 0
output_dir: ./Weyaxi-test

sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true

adapter: lora
lora_model_dir:

lora_r: 512
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
lora_modules_to_save:
  - embed_tokens
  - lm_head

wandb_project: huggingface 
wandb_entity:
wandb_watch:
wandb_run_id:
wandb_log_model:

hub_model_id: Weyaxi/Weyaxi-test

gradient_accumulation_steps: 4 # change
micro_batch_size: 2 # change
num_epochs: 3
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

warmup_steps: 10

save_steps: 20
save_total_limit: 5

debug:
#deepspeed: deepspeed/zero3_bf16.json
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
  eos_token: "<|im_end|>"
tokens:
  - "<|im_start|>"

Weyaxi-test

This model is a fine-tuned version of chargoddard/internlm2-20b-llama 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: bitsandbytes
load_in_8bit: True
load_in_4bit: False
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: fp4
bnb_4bit_use_double_quant: False
bnb_4bit_compute_dtype: float32

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0002
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_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

Framework versions

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

Weyaxi/Stellaris-internlm2-20b-r512

Weyaxi-test

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Weyaxi/
Stellaris-internlm2-20b-r512