See axolotl config

axolotl version: 0.3.0

base_model: ./models/deepseek-llm-67b-base
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer
is_llama_derived_model: true

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: train-all-4k-alpaca-deepseek.jsonl
    type: completion
dataset_prepared_path:
val_set_size: 0.0
output_dir: /workspace/volume/limarp-deepseek-qlora-out

adapter: qlora
lora_model_dir:

sequence_len: 4096
sample_packing: true
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: 70b-lora
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 2
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00015

train_on_inputs: true
group_by_length: false
bf16: true
fp16: false
tf32: true

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 10
evals_per_epoch:
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:

limarp-deepseek-67b-qlora

This model is an unofficial Deepseek 67B training on the LimaRP dataset.

Model description

Prompt format is the extended Alpaca format:

### Instruction:
Character's Persona: {bot character description}
User's Persona: {user character description}
Scenario: {what happens in the story}
Play the role of Character. You must engage in a roleplaying chat with User below this line. Do not write dialogues and narration for User.
### Input:
User: {utterance}
### Response:
Character: {utterance}
### Input:
User: {utterance}
### Response:
Character: {utterance}
(etc.)

Inspired by the previously named "Roleplay" preset in SillyTavern, with this version of LimaRP it is possible to append a length modifier to the response instruction sequence, like this:

### Input:
User: {utterance}

### Response: (length = medium)
Character: {utterance}

This has an immediately noticeable effect on bot responses. The lengths using during training are: micro, tiny, short, medium, long, massive, huge, enormous, humongous, unlimited. The recommended starting length is medium. Keep in mind that the AI can ramble or impersonate the user with very long messages.

The length control effect is reproducible, but the messages will not necessarily follow lengths very precisely, rather follow certain ranges on average, as seen in this table with data from tests made with one reply at the beginning of the conversation:

Response length control appears to work well also deep into the conversation. By omitting the modifier, the model will choose the most appropriate response length (although it might not necessarily be what the user desires).

Intended uses & limitations

The model will show biases similar to those observed in niche roleplaying forums on the Internet, besides those exhibited by the base model.

Training and evaluation data

For more details about LimaRP, see the dataset page.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.00015
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 8
• 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: 2

Framework versions

• Transformers 4.36.2
• Pytorch 2.0.1+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0

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

Framework versions

• PEFT 0.6.0