Instructions to use OsaurusAI/DeepSeek-V4-Flash-JANGTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use OsaurusAI/DeepSeek-V4-Flash-JANGTQ with Transformers:

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

pipe = pipeline("text-generation", model="OsaurusAI/DeepSeek-V4-Flash-JANGTQ")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("OsaurusAI/DeepSeek-V4-Flash-JANGTQ")
model = AutoModelForCausalLM.from_pretrained("OsaurusAI/DeepSeek-V4-Flash-JANGTQ")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use OsaurusAI/DeepSeek-V4-Flash-JANGTQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "OsaurusAI/DeepSeek-V4-Flash-JANGTQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "OsaurusAI/DeepSeek-V4-Flash-JANGTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/OsaurusAI/DeepSeek-V4-Flash-JANGTQ

SGLang

How to use OsaurusAI/DeepSeek-V4-Flash-JANGTQ 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 "OsaurusAI/DeepSeek-V4-Flash-JANGTQ" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "OsaurusAI/DeepSeek-V4-Flash-JANGTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "OsaurusAI/DeepSeek-V4-Flash-JANGTQ" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "OsaurusAI/DeepSeek-V4-Flash-JANGTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use OsaurusAI/DeepSeek-V4-Flash-JANGTQ with Docker Model Runner:
```
docker model run hf.co/OsaurusAI/DeepSeek-V4-Flash-JANGTQ
```

DeepSeek-V4-Flash-JANGTQ / config.json

Osaurus-AI

Upload config.json with huggingface_hub

b5c1f3e verified about 22 hours ago

raw

history blame contribute delete

2.23 kB

	{
	"architectures": [
	"DeepseekV4ForCausalLM"
	],
	"attention_bias": false,
	"attention_dropout": 0.0,
	"bos_token_id": 0,
	"eos_token_id": [
	1,
	128803
	],
	"expert_dtype": "fp4",
	"hc_eps": 1e-06,
	"hc_mult": 4,
	"hc_sinkhorn_iters": 20,
	"head_dim": 512,
	"hidden_act": "silu",
	"hidden_size": 4096,
	"index_head_dim": 128,
	"index_n_heads": 64,
	"index_topk": 512,
	"initializer_range": 0.02,
	"max_position_embeddings": 1048576,
	"model_type": "deepseek_v4",
	"moe_intermediate_size": 2048,
	"n_routed_experts": 256,
	"n_shared_experts": 1,
	"norm_topk_prob": true,
	"num_attention_heads": 64,
	"num_experts_per_tok": 6,
	"num_hidden_layers": 43,
	"num_hash_layers": 3,
	"num_key_value_heads": 1,
	"num_nextn_predict_layers": 1,
	"o_groups": 8,
	"o_lora_rank": 1024,
	"q_lora_rank": 1024,
	"qk_rope_head_dim": 64,
	"rms_norm_eps": 1e-06,
	"rope_theta": 10000,
	"routed_scaling_factor": 1.5,
	"scoring_func": "sqrtsoftplus",
	"sliding_window": 128,
	"swiglu_limit": 10.0,
	"tie_word_embeddings": false,
	"topk_method": "noaux_tc",
	"torch_dtype": "bfloat16",
	"transformers_version": "4.57.1",
	"use_cache": true,
	"vocab_size": 129280,
	"compress_rope_theta": 160000,
	"compress_ratios": [
	0,
	0,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	128,
	4,
	0
	],
	"rope_parameters": {
	"beta_fast": 32.0,
	"beta_slow": 1.0,
	"factor": 16.0,
	"original_max_position_embeddings": 65536,
	"rope_type": "yarn",
	"rope_theta": 10000.0
	},
	"quantization": {
	"bits": 8,
	"group_size": 32,
	"mode": "affine",
	"routed_expert_bits": 2,
	"mxtq_bits": {
	"routed_expert": 2,
	"attention": 8,
	"shared_expert": 8,
	"compressor": 8,
	"indexer": 8,
	"embed_tokens": 8,
	"lm_head": 8,
	"norms_router_hc": 16
	}
	},
	"routed_expert_bits": 2,
	"mxtq_seed": 42,
	"group_size": 32,
	"_name_or_path": "DSV4-Flash-JANGTQ_V3"
	}