148 lines
4.5 KiB
Markdown
148 lines
4.5 KiB
Markdown
---
|
|
license: apache-2.0
|
|
---
|
|
Based on Meta-Llama-3-8b-Instruct, and is governed by Meta Llama 3 License agreement:
|
|
https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct
|
|
|
|
|
|
We don't know how good this model is exactly in benchmarks since we have not benched this yet, but we think real prompts and usage is more telling anyways.
|
|
|
|
|
|
From our testing this model is:
|
|
|
|
- Less Refusals
|
|
- More Uncensored
|
|
- Follows requests better
|
|
- Can reply in requested formats better without adding unnecesary information
|
|
|
|
We are happy for anyone to try it out and give some feedback.
|
|
|
|
|
|
Training:
|
|
- 2048 sequence length, while the base model is 8192 sequence length. From testing it still performs the same 8192 context just fine.
|
|
- Trained on a modified and improved version of Cognitive Computations Eric Hartford's Dolphin dataset. https://huggingface.co/datasets/cognitivecomputations/dolphin
|
|
- Training duration is around 2 days on 2x RTX3090 on our own machine, using 4-bit loading and Qlora 64-rank 128-alpha resulting in ~2% trainable weights.
|
|
|
|
|
|
The goal for this model is to have the model less-censored and great at general tasks like the previous dolphin based models by Eric Hartford.
|
|
We started training this BEFORE they launched their own full weight trained Llama-3-8B-Dolphin-2.9 with their own curated datasets and the newer "Dolphin 2.9" dataset, but we think this model is still a unique take on Llama 3 8B Instruct and the dolphin dataset.
|
|
https://huggingface.co/cognitivecomputations/dolphin-2.9-llama3-8b
|
|
|
|
|
|
The difference with their dolphin 2.9 model is that we train this using Meta's new Llama 3 instruct format and not the regular ChatML format that Dolphin models are usually trained on.
|
|
This is because we think that it performed better using the format it was originally trained on.
|
|
|
|
Instruct format:
|
|
```
|
|
<|begin_of_text|><|start_header_id|>system<|end_header_id|>
|
|
|
|
{{ system_prompt }}<|eot_id|><|start_header_id|>user<|end_header_id|>
|
|
|
|
{{ user_message_1 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
|
|
|
|
{{ model_answer_1 }}<|eot_id|><|start_header_id|>user<|end_header_id|>
|
|
|
|
{{ user_message_2 }}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
|
|
```
|
|
|
|
|
|
Quants:
|
|
|
|
AWQ: https://huggingface.co/OwenArli/ArliAI-Llama-3-8B-Instruct-Dolfin-AWQ
|
|
|
|
GGUF: https://huggingface.co/OwenArli/ArliAI-Llama-3-8B-Instruct-Dolfin-v0.1-GGUF
|
|
|
|
FP16: https://huggingface.co/OwenArli/ArliAI-Llama-3-8B-Instruct-Dolfin
|
|
|
|
Exllamav2:
|
|
|
|
4bpw: https://huggingface.co/OwenArli/ArliAI-Llama-3-8B-Instruct-Dolfin-v0.1-exl2-h8-4bpw-exl2
|
|
|
|
8bpw: https://huggingface.co/OwenArli/ArliAI-Llama-3-8B-Instruct-Dolfin-v0.1-exl2-h8-8bpw-exl2
|
|
|
|
|
|
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
|
|
|
|
Axolotl Config:
|
|
```
|
|
base_model: Meta-Llama-3-8B-Instruct
|
|
model_type: LlamaForCausalLM
|
|
tokenizer_type: AutoTokenizer
|
|
|
|
train_on_inputs: false
|
|
group_by_length: false
|
|
load_in_8bit: false
|
|
load_in_4bit: true
|
|
strict: false
|
|
sequence_len: 2048
|
|
bf16: true
|
|
fp16: false
|
|
tf32: false
|
|
flash_attention: true
|
|
|
|
# Data
|
|
datasets:
|
|
- path: flan1m-universal-uncensored-system-2048.jsonl
|
|
type:
|
|
system_prompt: ""
|
|
system_format: "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\n{system}<|eot_id|><|start_header_id|>user<|end_header_id|>\n\n"
|
|
field_system: system
|
|
field_instruction: input
|
|
field_output: output
|
|
format: "{instruction}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n"
|
|
no_input_format: "{instruction}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n"
|
|
|
|
warmup_steps: 10
|
|
dataset_prepared_path: ./last_run_prepared
|
|
|
|
# Iterations
|
|
num_epochs: 1
|
|
saves_per_epoch: 4
|
|
|
|
# Evaluation
|
|
val_set_size: 0.01
|
|
eval_table_size:
|
|
eval_table_max_new_tokens:
|
|
eval_sample_packing: false
|
|
evals_per_epoch: 4
|
|
|
|
# LoRA
|
|
output_dir: ./qlora-out
|
|
adapter: qlora
|
|
lora_model_dir:
|
|
lora_r: 64
|
|
lora_alpha: 128
|
|
lora_dropout: 0.05
|
|
lora_target_linear: true
|
|
lora_fan_in_fan_out:
|
|
lora_target_modules:
|
|
save_safetensors: true
|
|
|
|
# Sampling
|
|
sample_packing: true
|
|
pad_to_sequence_len: true
|
|
|
|
# Batching
|
|
gradient_accumulation_steps: 32
|
|
micro_batch_size: 4
|
|
gradient_checkpointing: true
|
|
gradient_checkpointing_kwargs:
|
|
use_reentrant: true
|
|
|
|
# Optimizer
|
|
optimizer: paged_adamw_8bit
|
|
lr_scheduler: cosine
|
|
learning_rate: 0.0002
|
|
|
|
# Misc
|
|
early_stopping_patience:
|
|
resume_from_checkpoint:
|
|
logging_steps: 1
|
|
debug:
|
|
deepspeed: zero3_bf16.json
|
|
weight_decay: 0.1
|
|
special_tokens:
|
|
pad_token: <|end_of_text|>
|
|
```
|
|
|