python -m vllm.entrypoints.openai.api_server \
    --model macadeliccc/Samantha-Qwen2-7B-AWQ \
    --chat-template ./examples/template_chatml.jinja \
    --quantization awq

python

from openai import OpenAI
# Set OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://localhost:8000/v1"

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

chat_response = client.chat.completions.create(
    model="macadeliccc/Samantha-Qwen2-7B-AWQ",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Tell me a joke."},
    ]
)
print("Chat response:", chat_response)

Prompt Template

<|im_start|>system
You are  a friendly assistant.<|im_end|>
<|im_start|>user
What is the capital of France?<|im_end|>
<|im_start|>assistant
The capital of France is Paris.

Quants

See axolotl config

axolotl version: 0.4.0

yaml

base_model: Qwen/Qwen-7B
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

trust_remote_code: true

load_in_8bit: false
load_in_4bit: true
strict: false

datasets:
  - path: macadeliccc/opus_samantha
    type: sharegpt
    field: conversations
    conversation: chatml
  - path: uncensored-ultrachat.json
    type: sharegpt
    field: conversations
    conversation: chatml
  - path: openhermes_200k.json
    type: sharegpt
    field: conversations
    conversation: chatml
  - path: opus_instruct.json
    type: sharegpt
    field: conversations
    conversation: chatml

chat_template: chatml
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./outputs/lora-out

sequence_len: 2048
sample_packing: false
pad_to_sequence_len:

adapter: qlora
lora_model_dir:
lora_r: 32
lora_alpha: 16
lora_dropout: 0.05
lora_target_linear: true
lora_fan_in_fan_out:

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 1
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: false
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention:

warmup_steps: 250
evals_per_epoch: 4
eval_table_size:
eval_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens: