Instructions to use smohammadi/qat-llama-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use smohammadi/qat-llama-3B with Transformers:

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

pipe = pipeline("text-generation", model="smohammadi/qat-llama-3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("smohammadi/qat-llama-3B")
model = AutoModelForMultimodalLM.from_pretrained("smohammadi/qat-llama-3B")
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 smohammadi/qat-llama-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "smohammadi/qat-llama-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "smohammadi/qat-llama-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/smohammadi/qat-llama-3B

SGLang

How to use smohammadi/qat-llama-3B 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 "smohammadi/qat-llama-3B" \
    --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": "smohammadi/qat-llama-3B",
		"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 "smohammadi/qat-llama-3B" \
        --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": "smohammadi/qat-llama-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use smohammadi/qat-llama-3B with Docker Model Runner:
```
docker model run hf.co/smohammadi/qat-llama-3B
```

qat-llama-3B / README.md

smohammadi

Model save

9028a1c verified 10 months ago

preview code

raw

history blame contribute delete

3.43 kB

metadata

library_name: transformers
license: llama3.2
base_model: meta-llama/Llama-3.2-3B-Instruct
tags:
  - axolotl
  - generated_from_trainer
datasets:
  - yahma/alpaca-cleaned
model-index:
  - name: qat-llama-3B
    results: []

See axolotl config

axolotl version: 0.13.0.dev0

base_model: meta-llama/Llama-3.2-3B-Instruct
# Automatically upload checkpoint and final model to HF
hub_model_id: smohammadi/qat-llama-3B # username/custom_model_name

load_in_8bit: false
load_in_4bit: false
strict: false


    #liger_rope: true
    #liger_rms_norm: true
    #liger_glu_activation: true
    #liger_layer_norm: true
    #
    #liger_fused_linear_cross_entropy: true

datasets:
  - path: yahma/alpaca-cleaned
    type: alpaca
    split: train[:95%]
output_dir: ./outputs/qat-train_on_inputs/
dataset_prepared_path: ./outputs/ds_prepared_new_token
  #sample_packing: true

sequence_len: 8192
flash_attention: true
  #flex_attention: true
  #flex_attn_compile_kwargs:
  # dynamic: false
  # mode: max-autotune-no-cudagraphs

qat:
   activation_dtype: int8
   weight_dtype: int4
   group_size: 32

wandb_project: qat_v2
wandb_entity:
wandb_watch:
wandb_name: qat-train-on-inputs
wandb_log_model:

gradient_accumulation_steps: 1
micro_batch_size: 16
num_epochs: 1
optimizer: adamw_torch_fused
train_on_inputs: true

  #cosine_constant_lr_ratio: 0
  #cosine_min_lr_ratio: 1.0
lr_scheduler: constant
learning_rate: 2e-5
save_only_model: true
bf16: true

resume_from_checkpoint:
logging_steps: 1
include_tkps: true
evals_per_epoch: 1
saves_per_epoch: 1

  #warmup_ratio: 0.1
weight_decay: 0.0

fsdp_config:
  fsdp_version: 2
  fsdp_offload_params: false
  fsdp_cpu_ram_efficient_loading: False
  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
  fsdp_state_dict_type: FULL_STATE_DICT
  fsdp_sharding_strategy: FULL_SHARD
  fsdp_reshard_after_forward: true
  fsdp_activation_checkpointing: true

special_tokens:
  pad_token: <|finetune_right_pad_id|>

qat-llama-3B

This model is a fine-tuned version of meta-llama/Llama-3.2-3B-Instruct on the yahma/alpaca-cleaned dataset.

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: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
distributed_type: multi-GPU
num_devices: 4
total_train_batch_size: 64
total_eval_batch_size: 64
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: constant
lr_scheduler_warmup_steps: 23
training_steps: 769

Training results

Framework versions

Transformers 4.55.3
Pytorch 2.7.1+cu126
Datasets 4.0.0
Tokenizers 0.21.4