Instructions to use Flubstar/LongCat-2.0 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Flubstar/LongCat-2.0 with Transformers:

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

pipe = pipeline("text-generation", model="Flubstar/LongCat-2.0")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import LongcatCausalLM
model = LongcatCausalLM.from_pretrained("Flubstar/LongCat-2.0", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Flubstar/LongCat-2.0 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Flubstar/LongCat-2.0"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Flubstar/LongCat-2.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Flubstar/LongCat-2.0

SGLang

How to use Flubstar/LongCat-2.0 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 "Flubstar/LongCat-2.0" \
    --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": "Flubstar/LongCat-2.0",
		"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 "Flubstar/LongCat-2.0" \
        --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": "Flubstar/LongCat-2.0",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Flubstar/LongCat-2.0 with Docker Model Runner:
```
docker model run hf.co/Flubstar/LongCat-2.0
```

LongCat-2.0 / config.json

Flubstar

Duplicate from meituan-longcat/LongCat-2.0

fd9b8d8 1 day ago

Raw

History Blame Contribute Delete

1.52 kB

	{
	"architectures": [
	"LongcatCausalLM"
	],
	"attention_bias": false,
	"attention_dropout": 0.0,
	"vocab_size": 163840,
	"hidden_size": 8192,
	"ffn_hidden_size": 12288,
	"expert_ffn_hidden_size": 2048,
	"num_layers": 38,
	"num_attention_heads": 64,
	"kv_lora_rank": 512,
	"q_lora_rank": 1536,
	"qk_rope_head_dim": 64,
	"v_head_dim": 128,
	"qk_nope_head_dim": 128,
	"mla_scale_q_lora": true,
	"mla_scale_kv_lora": true,
	"routed_scaling_factor": 9,
	"n_routed_experts": 768,
	"max_position_embeddings": 262144,
	"rms_norm_eps": 1e-5,
	"use_cache": true,
	"bos_token_id": 1,
	"eos_token_id": 2,
	"rope_theta": 1000000.0,
	"rope_scaling": {
	"original_max_position_embeddings": 8192,
	"rope_type": "deepseek_yarn",
	"factor": 120,
	"beta_fast": 32,
	"beta_slow": 1,
	"mscale": 1,
	"mscale_all_dim": 1
	},
	"attention_method": "MLA",
	"zero_expert_num": 128,
	"zero_expert_type": "identity",
	"moe_topk": 12,
	"use_mla": 1,
	"moe_switch_token_num": 1024,
	"moe_impl": "mix",
	"oe_vocab_size_ratio": 100.567,
	"oe_neighbor_num": 5,
	"oe_split_num": 4,

	"mtp_num_layers": 3,
	"mtp_replicate_modules": true,
	"mtp_disable_over_tokenizer": true,

	"index_n_heads": 32,
	"index_head_dim": 128,
	"index_topk": 2048,
	"index_k_norm_type": "rms",
	"cli_factor": 2,
	"dsa_mtp_cli": true,
	"index_local_tokens": 1024,
	"index_init_tokens": 16
	}