Instructions to use kennylam/Breeze-7B-Cantonese-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kennylam/Breeze-7B-Cantonese-v0.1 with Transformers:

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

pipe = pipeline("text-generation", model="kennylam/Breeze-7B-Cantonese-v0.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kennylam/Breeze-7B-Cantonese-v0.1")
model = AutoModelForCausalLM.from_pretrained("kennylam/Breeze-7B-Cantonese-v0.1", 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 kennylam/Breeze-7B-Cantonese-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kennylam/Breeze-7B-Cantonese-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kennylam/Breeze-7B-Cantonese-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/kennylam/Breeze-7B-Cantonese-v0.1

SGLang

How to use kennylam/Breeze-7B-Cantonese-v0.1 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 "kennylam/Breeze-7B-Cantonese-v0.1" \
    --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": "kennylam/Breeze-7B-Cantonese-v0.1",
		"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 "kennylam/Breeze-7B-Cantonese-v0.1" \
        --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": "kennylam/Breeze-7B-Cantonese-v0.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use kennylam/Breeze-7B-Cantonese-v0.1 with Docker Model Runner:
```
docker model run hf.co/kennylam/Breeze-7B-Cantonese-v0.1
```

Breeze-7B-Cantonese-v0.1 / axolotl-config.yml

kennylam

First commit for axolotl-config.yml

0e08b7b over 2 years ago

Raw

History Blame Contribute Delete

2.97 kB

	base_model: MediaTek-Research/Breeze-7B-Base-v0_1
	model_type: MistralForCausalLM
	tokenizer_type: LlamaTokenizer
	is_mistral_derived_model: true

	load_in_8bit: false
	load_in_4bit: true
	strict: false


	datasets:
	- path: hon9kon9ize/yue-alpaca
	type: alpaca
	- path: indiejoseph/wikipedia-translate-zhhk-zhcn
	type:
	system_prompt: ""
	field_instruction: zh
	field_output: yue
	format: \|-
	[INST]
	翻譯下面中文至粵語廣東話(Cantonese)。

	{instruction}
	[/INST]
	- path: indiejoseph/wikipedia-zh-yue-summaries
	type:
	system_prompt: ""
	field_instruction: content
	field_output: summary
	format: \|-
	[INST]
	用粵語廣東話(Cantonese)總結一吓。

	{instruction}
	[/INST]
	- path: indiejoseph/wikipedia-zh-yue-summaries
	type:
	system_prompt: ""
	field_instruction: content
	field_output: title
	format: \|-
	[INST]
	粵語廣東話(Cantonese), 呢篇嘢主題係咩?

	{instruction}
	[/INST]
	- path: indiejoseph/wikipedia-zh-yue-summaries
	type:
	system_prompt: ""
	field_instruction: content
	field_output: category
	format: \|-
	[INST]
	粵語廣東話(Cantonese), 呢篇嘢講緊咩? 係咩分類?

	{instruction}
	[/INST]
	- path: indiejoseph/wikipedia-zh-yue-qa
	type:
	system_prompt: ""
	field_instruction: question
	field_system: title
	field_output: answer
	format: \|-
	[INST]
	粵語廣東話(Cantonese), 以下係關於「{system}」嘅問題。

	{instruction}
	[/INST]

	dataset_prepared_path: last_run_prepared
	val_set_size: 0.05
	output_dir: loras/Breeze-7B-Cantonese-v0.1
	save_safetensors: true
	#eval_sample_packing: False

	## You can optionally freeze the entire model and unfreeze a subset of parameters
	unfrozen_parameters:
	# - lm_head.*
	# - model.embed_tokens.*
	# - model.layers.2[0-9]+.block_sparse_moe.gate.*
	# - model.layers.2[0-9]+.block_sparse_moe.experts.*
	# - model.layers.3[0-9]+.block_sparse_moe.gate.*
	# - model.layers.3[0-9]+.block_sparse_moe.experts.*

	model_config:
	output_router_logits: true

	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_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: true
	fp16: false
	tf32: false

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

	loss_watchdog_threshold: 5.0
	loss_watchdog_patience: 3

	warmup_steps: 10
	evals_per_epoch: 4
	eval_table_size:
	eval_table_max_new_tokens: 128
	saves_per_epoch: 1
	debug:
	deepspeed:
	weight_decay: 0.0
	fsdp:
	fsdp_config:
	special_tokens: