YoungPanda/up2chatml

See axolotl config

axolotl version: 0.4.0

base_model: upstage/SOLAR-10.7B-v1.0
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
is_llama_derived_model: true
trust_remote_code: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: Drewskidang/share
    type: sharegpt
    split: train
    conversation: chatml
  - path: Drewskidang/shareinstruct
    type: sharegpt
    split: train
    conversation: chatml

dataset_prepared_path: last_run_prepared
val_set_size: 0.05
output_dir: ./chatml

sequence_len: 4096
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false

adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 12
num_epochs: 5
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.0002

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

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
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: 100
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed: #deepspeed_configs/zero2.json # multi-gpu only
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
tokens: # these are delimiters
  - "<|im_start|>"
  - "<|im_end|>"

chatml

This model is a fine-tuned version of upstage/SOLAR-10.7B-v1.0 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.1262

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: 0.0002
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 4
• total_train_batch_size: 384
• total_eval_batch_size: 96
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss
1.4068	0.31	1	1.4339
1.4144	0.62	2	1.3014
1.2489	0.92	3	1.2079
1.1695	1.15	4	1.2505
1.2222	1.46	5	1.2246
1.1716	1.77	6	1.5272
1.4521	2.08	7	1.2458
1.1637	2.31	8	1.2050
1.118	2.62	9	1.3758
1.2671	2.92	10	1.2687
1.1615	3.15	11	1.1509
0.9979	3.46	12	1.2254
1.0704	3.77	13	1.1332
0.9714	4.08	14	1.1505
0.9191	4.31	15	1.1262

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.17.0
• Tokenizers 0.15.0