Instructions to use vijil/llama3-8b-instruct-consistent_sft-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use vijil/llama3-8b-instruct-consistent_sft-v2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="vijil/llama3-8b-instruct-consistent_sft-v2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("vijil/llama3-8b-instruct-consistent_sft-v2")
model = AutoModelForCausalLM.from_pretrained("vijil/llama3-8b-instruct-consistent_sft-v2", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use vijil/llama3-8b-instruct-consistent_sft-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "vijil/llama3-8b-instruct-consistent_sft-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vijil/llama3-8b-instruct-consistent_sft-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/vijil/llama3-8b-instruct-consistent_sft-v2

SGLang

How to use vijil/llama3-8b-instruct-consistent_sft-v2 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "vijil/llama3-8b-instruct-consistent_sft-v2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vijil/llama3-8b-instruct-consistent_sft-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "vijil/llama3-8b-instruct-consistent_sft-v2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "vijil/llama3-8b-instruct-consistent_sft-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use vijil/llama3-8b-instruct-consistent_sft-v2 with Docker Model Runner:
```
docker model run hf.co/vijil/llama3-8b-instruct-consistent_sft-v2
```

See axolotl config

axolotl version: 0.4.0

base_model: meta-llama/Meta-Llama-3-8B-Instruct
model_type: LlamaForCausalLM
tokenizer_type: AutoTokenizer

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: ./my_data/consistency_finetune-data-v2-axolotl_fft.jsonl
    type: alpaca
dataset_prepared_path: last_run_prepared
val_set_size: 0.05
eval_sample_packing: False
output_dir: ./consistency_finetune-data-v2-axolotl_fft
hub_model_id: vijil/llama3-8b-instruct-consistent_sft-v2

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 1
num_epochs: 4
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.00005

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
flash_attn_cross_entropy: false
flash_attn_rms_norm: true
flash_attn_fuse_qkv: false
flash_attn_fuse_mlp: true

warmup_ratio: 0.02 
evals_per_epoch: 2
eval_table_size:
saves_per_epoch: 1
debug:
deepspeed: deepspeed_configs/zero2.json # multi-gpu only
weight_decay: 0.1
adam_beta1: 0.9
adam_beta2: 0.95
adam_epsilon: 0.00000001
max_grad_norm: 1.0
fsdp:
fsdp_config:
special_tokens:
  pad_token: <|end_of_text|>

llama3-8b-instruct-consistent_sft-v2

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.2056

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-05
train_batch_size: 1
eval_batch_size: 1
seed: 42
distributed_type: multi-GPU
num_devices: 10
gradient_accumulation_steps: 8
total_train_batch_size: 80
total_eval_batch_size: 10
optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-08
lr_scheduler_type: cosine
num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss
3.1682	0.09	1	3.1898
0.9665	0.54	6	0.8419
0.4173	1.06	12	0.4193
0.2811	1.6	18	0.3025
0.1379	2.1	24	0.2382
0.1176	2.64	30	0.2314
0.075	3.16	36	0.2094
0.066	3.7	42	0.2056

Framework versions

Transformers 4.40.0.dev0
Pytorch 2.2.1
Datasets 2.15.0
Tokenizers 0.15.0

Downloads last month: 11

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for vijil/llama3-8b-instruct-consistent_sft-v2

Base model

meta-llama/Meta-Llama-3-8B-Instruct

Finetuned

(1076)

this model