Instructions to use tiny-random/bailing-moe-v2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tiny-random/bailing-moe-v2.5 with Transformers:

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

pipe = pipeline("text-generation", model="tiny-random/bailing-moe-v2.5", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("tiny-random/bailing-moe-v2.5", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use tiny-random/bailing-moe-v2.5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tiny-random/bailing-moe-v2.5"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tiny-random/bailing-moe-v2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tiny-random/bailing-moe-v2.5

SGLang

How to use tiny-random/bailing-moe-v2.5 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 "tiny-random/bailing-moe-v2.5" \
    --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": "tiny-random/bailing-moe-v2.5",
		"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 "tiny-random/bailing-moe-v2.5" \
        --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": "tiny-random/bailing-moe-v2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tiny-random/bailing-moe-v2.5 with Docker Model Runner:
```
docker model run hf.co/tiny-random/bailing-moe-v2.5
```

bailing-moe-v2.5 / README.md

yujiepan

Upload folder using huggingface_hub

bbdb4c5 verified 3 months ago

preview code

raw

history blame contribute delete

6.58 kB

	---
	library_name: transformers
	base_model:
	- inclusionAI/Ring-2.5-1T
	---

	This tiny model is intended for debugging. It is randomly initialized using the configuration adapted from [inclusionAI/Ring-2.5-1T](https://huggingface.co/inclusionAI/Ring-2.5-1T).

	\| File path \| Size \|
	\|------\|------\|
	\| model.safetensors \| 6.4MB \|


	### Example usage:

	```python
	import torch
	import transformers
	from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
	transformers.utils.import_utils.is_torch_fx_available = transformers.utils.import_utils.is_torch_available

	model_id = "tiny-random/bailing-moe-v2.5"
	tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
	model = AutoModelForCausalLM.from_pretrained(
	model_id,
	torch_dtype=torch.bfloat16,
	trust_remote_code=True,
	)
	pipe = pipeline('text-generation', model=model, tokenizer=tokenizer, trust_remote_code=True)
	print(pipe('Write an article about Artificial Intelligence.', max_new_tokens=16))
	```

	### Codes to create this repo:

	<details>
	<summary>Click to expand</summary>

	```python
	import json
	from pathlib import Path

	import accelerate
	import torch
	import transformers
	from huggingface_hub import file_exists, hf_hub_download
	from transformers import (
	AutoConfig,
	AutoModelForCausalLM,
	AutoTokenizer,
	GenerationConfig,
	set_seed,
	)
	transformers.utils.import_utils.is_torch_fx_available = transformers.utils.import_utils.is_torch_available
	source_model_id = "inclusionAI/Ring-2.5-1T"
	save_folder = "/tmp/tiny-random/bailing-moe-v25"

	processor = AutoTokenizer.from_pretrained(source_model_id, trust_remote_code=True)
	processor.save_pretrained(save_folder)

	with open(hf_hub_download(source_model_id, filename='config.json', repo_type='model'), 'r', encoding='utf-8') as f:
	config_json = json.load(f)
	for k, v in config_json['auto_map'].items():
	config_json['auto_map'][k] = f'{source_model_id}--{v}'

	# config_json['head_dim'] = 32
	config_json['hidden_size'] = 8
	config_json['intermediate_size'] = 32
	config_json['moe_intermediate_size'] = 32
	config_json['moe_shared_expert_intermediate_size'] = 32
	config_json['first_k_dense_replace'] = 1
	config_json['num_attention_heads'] = 4
	config_json['num_hidden_layers'] = 2
	config_json['num_key_value_heads'] = 4
	config_json['q_lora_rank'] = 32
	config_json['layer_group_size'] = 2
	del config_json['quantization_config']

	with open(f"{save_folder}/config.json", "w", encoding='utf-8') as f:
	json.dump(config_json, f, indent=2)

	config = AutoConfig.from_pretrained(
	save_folder,
	trust_remote_code=True,
	)
	print(config)
	automap = config_json['auto_map']
	torch.set_default_dtype(torch.bfloat16)
	model = AutoModelForCausalLM.from_config(config, trust_remote_code=True)
	torch.set_default_dtype(torch.float32)

	if file_exists(filename="generation_config.json", repo_id=source_model_id, repo_type='model'):
	model.generation_config = GenerationConfig.from_pretrained(
	source_model_id, trust_remote_code=True,
	)
	set_seed(42)
	model = model.cpu()
	with torch.no_grad():
	for name, p in sorted(model.named_parameters()):
	torch.nn.init.normal_(p, 0, 0.1)
	print(name, p.shape)
	model.model.layers[1].mlp.gate.expert_bias = model.model.layers[1].mlp.gate.expert_bias.float()
	model.save_pretrained(save_folder)
	print(model)
	with open(f"{save_folder}/config.json", "r", encoding='utf-8') as f:
	config_json = json.load(f)
	config_json['auto_map'] = automap
	with open(f"{save_folder}/config.json", "w", encoding='utf-8') as f:
	json.dump(config_json, f, indent=2)
	for python_file in Path(save_folder).glob('*.py'):
	python_file.unlink()
	```

	</details>

	### Printing the model:

	<details><summary>Click to expand</summary>

	```text
	BailingMoeV2_5ForCausalLM(
	(model): BailingMoeV2_5Model(
	(word_embeddings): Embedding(157184, 8, padding_idx=156892)
	(layers): ModuleList(
	(0): BailingMoeV2_5DecoderLayer(
	(attention): BailingMoeV2_5LinearAttention(
	(query_key_value): Linear(in_features=8, out_features=1536, bias=False)
	(query_layernorm): BailingMoeV2_5RMSNorm()
	(key_layernorm): BailingMoeV2_5RMSNorm()
	(rotary_emb): BailingMoeV2_5RotaryEmbedding()
	(dense): Linear(in_features=512, out_features=8, bias=False)
	(g_proj): Linear(in_features=8, out_features=512, bias=False)
	(g_norm): BailingMoeV2_5GroupRMSNorm()
	)
	(mlp): BailingMoeV2_5MLP(
	(gate_proj): Linear(in_features=8, out_features=32, bias=False)
	(up_proj): Linear(in_features=8, out_features=32, bias=False)
	(down_proj): Linear(in_features=32, out_features=8, bias=False)
	(act_fn): SiLUActivation()
	)
	(input_layernorm): BailingMoeV2_5RMSNorm()
	(post_attention_layernorm): BailingMoeV2_5RMSNorm()
	)
	(1): BailingMoeV2_5DecoderLayer(
	(attention): BailingMoeV2_5MultiLatentAttention(
	(q_a_proj): Linear(in_features=8, out_features=32, bias=False)
	(q_a_layernorm): BailingMoeV2_5RMSNorm()
	(q_b_proj): Linear(in_features=32, out_features=768, bias=False)
	(kv_a_proj_with_mqa): Linear(in_features=8, out_features=576, bias=False)
	(kv_a_layernorm): BailingMoeV2_5RMSNorm()
	(kv_b_proj): Linear(in_features=512, out_features=1024, bias=False)
	(dense): Linear(in_features=512, out_features=8, bias=False)
	)
	(mlp): BailingMoeV2_5SparseMoeBlock(
	(experts): ModuleList(
	(0-255): 256 x BailingMoeV2_5MLP(
	(gate_proj): Linear(in_features=8, out_features=32, bias=False)
	(up_proj): Linear(in_features=8, out_features=32, bias=False)
	(down_proj): Linear(in_features=32, out_features=8, bias=False)
	(act_fn): SiLUActivation()
	)
	)
	(gate): BailingMoeV2_5Gate()
	(shared_experts): BailingMoeV2_5MLP(
	(gate_proj): Linear(in_features=8, out_features=32, bias=False)
	(up_proj): Linear(in_features=8, out_features=32, bias=False)
	(down_proj): Linear(in_features=32, out_features=8, bias=False)
	(act_fn): SiLUActivation()
	)
	)
	(input_layernorm): BailingMoeV2_5RMSNorm()
	(post_attention_layernorm): BailingMoeV2_5RMSNorm()
	)
	)
	(norm): BailingMoeV2_5RMSNorm()
	(rotary_emb): BailingMoeV2_5RotaryEmbedding()
	(rotary_emb_mla): BailingMoeV2_5MLARotaryEmbedding()
	)
	(lm_head): Linear(in_features=8, out_features=157184, bias=False)
	)
	```

	</details>