Instructions to use RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8") model = AutoModelForCausalLM.from_pretrained("RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8
- SGLang
How to use RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8 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 "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8" \ --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": "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8", "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 "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8" \ --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": "RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8 with Docker Model Runner:
docker model run hf.co/RedHatAI/DeepSeek-V4-Pro-NVFP4-FP8
| { | |
| "architectures": [ | |
| "DeepseekV4ForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 0, | |
| "compress_ratios": [ | |
| 128, | |
| 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, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 128, | |
| 4, | |
| 0 | |
| ], | |
| "compress_rope_theta": 160000, | |
| "eos_token_id": 1, | |
| "expert_dtype": "fp4", | |
| "hc_eps": 1e-06, | |
| "hc_mult": 4, | |
| "hc_sinkhorn_iters": 20, | |
| "head_dim": 512, | |
| "hidden_act": "silu", | |
| "hidden_size": 7168, | |
| "index_head_dim": 128, | |
| "index_n_heads": 64, | |
| "index_topk": 1024, | |
| "initializer_range": 0.02, | |
| "max_position_embeddings": 1048576, | |
| "model_type": "deepseek_v4", | |
| "moe_intermediate_size": 3072, | |
| "n_routed_experts": 384, | |
| "n_shared_experts": 1, | |
| "norm_topk_prob": true, | |
| "num_attention_heads": 128, | |
| "num_experts_per_tok": 6, | |
| "num_hash_layers": 3, | |
| "num_hidden_layers": 61, | |
| "num_key_value_heads": 1, | |
| "num_nextn_predict_layers": 1, | |
| "o_groups": 16, | |
| "o_lora_rank": 1024, | |
| "q_lora_rank": 1536, | |
| "qk_rope_head_dim": 64, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": { | |
| "beta_fast": 32, | |
| "beta_slow": 1, | |
| "factor": 16, | |
| "original_max_position_embeddings": 65536, | |
| "type": "yarn" | |
| }, | |
| "rope_theta": 10000, | |
| "routed_scaling_factor": 2.5, | |
| "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, | |
| "quantization_config": { | |
| "config_groups": { | |
| "group_0": { | |
| "format": "float-quantized", | |
| "input_activations": { | |
| "actorder": null, | |
| "block_structure": null, | |
| "dynamic": true, | |
| "group_size": 128, | |
| "num_bits": 8, | |
| "observer": null, | |
| "observer_kwargs": {}, | |
| "scale_dtype": null, | |
| "strategy": "group", | |
| "symmetric": true, | |
| "type": "float", | |
| "zp_dtype": null | |
| }, | |
| "output_activations": null, | |
| "targets": [ | |
| "re:.*attn.*(fused_wqa_wkv|wq_b|wo_a|wo_b)$" | |
| ], | |
| "weights": { | |
| "actorder": null, | |
| "block_structure": [ | |
| 128, | |
| 128 | |
| ], | |
| "dynamic": false, | |
| "group_size": null, | |
| "num_bits": 8, | |
| "observer": "memoryless_minmax", | |
| "observer_kwargs": {}, | |
| "scale_dtype": null, | |
| "strategy": "block", | |
| "symmetric": true, | |
| "type": "float", | |
| "zp_dtype": null | |
| } | |
| }, | |
| "group_1": { | |
| "format": "nvfp4-pack-quantized", | |
| "input_activations": { | |
| "actorder": null, | |
| "block_structure": null, | |
| "dynamic": "local", | |
| "group_size": 16, | |
| "num_bits": 4, | |
| "observer": "static_minmax", | |
| "observer_kwargs": {}, | |
| "scale_dtype": "torch.float8_e4m3fn", | |
| "strategy": "tensor_group", | |
| "symmetric": true, | |
| "type": "float", | |
| "zp_dtype": null | |
| }, | |
| "output_activations": null, | |
| "targets": [ | |
| "re:.*ffn.*" | |
| ], | |
| "weights": { | |
| "actorder": null, | |
| "block_structure": null, | |
| "dynamic": false, | |
| "group_size": 16, | |
| "num_bits": 4, | |
| "observer": "memoryless_minmax", | |
| "observer_kwargs": {}, | |
| "scale_dtype": "torch.float8_e4m3fn", | |
| "strategy": "tensor_group", | |
| "symmetric": true, | |
| "type": "float", | |
| "zp_dtype": null | |
| } | |
| } | |
| }, | |
| "format": "mixed-precision", | |
| "global_compression_ratio": null, | |
| "ignore": [ | |
| "model.layers.0.self_attn.compressor.kv_proj", | |
| "model.layers.0.self_attn.compressor.gate_proj", | |
| "model.layers.0.mlp.gate", | |
| "model.layers.1.self_attn.compressor.kv_proj", | |
| "model.layers.1.self_attn.compressor.gate_proj", | |
| "model.layers.1.mlp.gate", | |
| "model.layers.2.self_attn.compressor.kv_proj", | |
| "model.layers.2.self_attn.compressor.gate_proj", | |
| "model.layers.2.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.2.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.2.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.2.mlp.gate", | |
| "model.layers.3.self_attn.compressor.kv_proj", | |
| "model.layers.3.self_attn.compressor.gate_proj", | |
| "model.layers.3.mlp.gate", | |
| "model.layers.4.self_attn.compressor.kv_proj", | |
| "model.layers.4.self_attn.compressor.gate_proj", | |
| "model.layers.4.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.4.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.4.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.4.mlp.gate", | |
| "model.layers.5.self_attn.compressor.kv_proj", | |
| "model.layers.5.self_attn.compressor.gate_proj", | |
| "model.layers.5.mlp.gate", | |
| "model.layers.6.self_attn.compressor.kv_proj", | |
| "model.layers.6.self_attn.compressor.gate_proj", | |
| "model.layers.6.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.6.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.6.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.6.mlp.gate", | |
| "model.layers.7.self_attn.compressor.kv_proj", | |
| "model.layers.7.self_attn.compressor.gate_proj", | |
| "model.layers.7.mlp.gate", | |
| "model.layers.8.self_attn.compressor.kv_proj", | |
| "model.layers.8.self_attn.compressor.gate_proj", | |
| "model.layers.8.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.8.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.8.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.8.mlp.gate", | |
| "model.layers.9.self_attn.compressor.kv_proj", | |
| "model.layers.9.self_attn.compressor.gate_proj", | |
| "model.layers.9.mlp.gate", | |
| "model.layers.10.self_attn.compressor.kv_proj", | |
| "model.layers.10.self_attn.compressor.gate_proj", | |
| "model.layers.10.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.10.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.10.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.10.mlp.gate", | |
| "model.layers.11.self_attn.compressor.kv_proj", | |
| "model.layers.11.self_attn.compressor.gate_proj", | |
| "model.layers.11.mlp.gate", | |
| "model.layers.12.self_attn.compressor.kv_proj", | |
| "model.layers.12.self_attn.compressor.gate_proj", | |
| "model.layers.12.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.12.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.12.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.12.mlp.gate", | |
| "model.layers.13.self_attn.compressor.kv_proj", | |
| "model.layers.13.self_attn.compressor.gate_proj", | |
| "model.layers.13.mlp.gate", | |
| "model.layers.14.self_attn.compressor.kv_proj", | |
| "model.layers.14.self_attn.compressor.gate_proj", | |
| "model.layers.14.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.14.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.14.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.14.mlp.gate", | |
| "model.layers.15.self_attn.compressor.kv_proj", | |
| "model.layers.15.self_attn.compressor.gate_proj", | |
| "model.layers.15.mlp.gate", | |
| "model.layers.16.self_attn.compressor.kv_proj", | |
| "model.layers.16.self_attn.compressor.gate_proj", | |
| "model.layers.16.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.16.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.16.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.16.mlp.gate", | |
| "model.layers.17.self_attn.compressor.kv_proj", | |
| "model.layers.17.self_attn.compressor.gate_proj", | |
| "model.layers.17.mlp.gate", | |
| "model.layers.18.self_attn.compressor.kv_proj", | |
| "model.layers.18.self_attn.compressor.gate_proj", | |
| "model.layers.18.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.18.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.18.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.18.mlp.gate", | |
| "model.layers.19.self_attn.compressor.kv_proj", | |
| "model.layers.19.self_attn.compressor.gate_proj", | |
| "model.layers.19.mlp.gate", | |
| "model.layers.20.self_attn.compressor.kv_proj", | |
| "model.layers.20.self_attn.compressor.gate_proj", | |
| "model.layers.20.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.20.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.20.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.20.mlp.gate", | |
| "model.layers.21.self_attn.compressor.kv_proj", | |
| "model.layers.21.self_attn.compressor.gate_proj", | |
| "model.layers.21.mlp.gate", | |
| "model.layers.22.self_attn.compressor.kv_proj", | |
| "model.layers.22.self_attn.compressor.gate_proj", | |
| "model.layers.22.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.22.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.22.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.22.mlp.gate", | |
| "model.layers.23.self_attn.compressor.kv_proj", | |
| "model.layers.23.self_attn.compressor.gate_proj", | |
| "model.layers.23.mlp.gate", | |
| "model.layers.24.self_attn.compressor.kv_proj", | |
| "model.layers.24.self_attn.compressor.gate_proj", | |
| "model.layers.24.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.24.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.24.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.24.mlp.gate", | |
| "model.layers.25.self_attn.compressor.kv_proj", | |
| "model.layers.25.self_attn.compressor.gate_proj", | |
| "model.layers.25.mlp.gate", | |
| "model.layers.26.self_attn.compressor.kv_proj", | |
| "model.layers.26.self_attn.compressor.gate_proj", | |
| "model.layers.26.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.26.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.26.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.26.mlp.gate", | |
| "model.layers.27.self_attn.compressor.kv_proj", | |
| "model.layers.27.self_attn.compressor.gate_proj", | |
| "model.layers.27.mlp.gate", | |
| "model.layers.28.self_attn.compressor.kv_proj", | |
| "model.layers.28.self_attn.compressor.gate_proj", | |
| "model.layers.28.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.28.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.28.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.28.mlp.gate", | |
| "model.layers.29.self_attn.compressor.kv_proj", | |
| "model.layers.29.self_attn.compressor.gate_proj", | |
| "model.layers.29.mlp.gate", | |
| "model.layers.30.self_attn.compressor.kv_proj", | |
| "model.layers.30.self_attn.compressor.gate_proj", | |
| "model.layers.30.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.30.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.30.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.30.mlp.gate", | |
| "model.layers.31.self_attn.compressor.kv_proj", | |
| "model.layers.31.self_attn.compressor.gate_proj", | |
| "model.layers.31.mlp.gate", | |
| "model.layers.32.self_attn.compressor.kv_proj", | |
| "model.layers.32.self_attn.compressor.gate_proj", | |
| "model.layers.32.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.32.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.32.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.32.mlp.gate", | |
| "model.layers.33.self_attn.compressor.kv_proj", | |
| "model.layers.33.self_attn.compressor.gate_proj", | |
| "model.layers.33.mlp.gate", | |
| "model.layers.34.self_attn.compressor.kv_proj", | |
| "model.layers.34.self_attn.compressor.gate_proj", | |
| "model.layers.34.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.34.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.34.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.34.mlp.gate", | |
| "model.layers.35.self_attn.compressor.kv_proj", | |
| "model.layers.35.self_attn.compressor.gate_proj", | |
| "model.layers.35.mlp.gate", | |
| "model.layers.36.self_attn.compressor.kv_proj", | |
| "model.layers.36.self_attn.compressor.gate_proj", | |
| "model.layers.36.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.36.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.36.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.36.mlp.gate", | |
| "model.layers.37.self_attn.compressor.kv_proj", | |
| "model.layers.37.self_attn.compressor.gate_proj", | |
| "model.layers.37.mlp.gate", | |
| "model.layers.38.self_attn.compressor.kv_proj", | |
| "model.layers.38.self_attn.compressor.gate_proj", | |
| "model.layers.38.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.38.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.38.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.38.mlp.gate", | |
| "model.layers.39.self_attn.compressor.kv_proj", | |
| "model.layers.39.self_attn.compressor.gate_proj", | |
| "model.layers.39.mlp.gate", | |
| "model.layers.40.self_attn.compressor.kv_proj", | |
| "model.layers.40.self_attn.compressor.gate_proj", | |
| "model.layers.40.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.40.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.40.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.40.mlp.gate", | |
| "model.layers.41.self_attn.compressor.kv_proj", | |
| "model.layers.41.self_attn.compressor.gate_proj", | |
| "model.layers.41.mlp.gate", | |
| "model.layers.42.self_attn.compressor.kv_proj", | |
| "model.layers.42.self_attn.compressor.gate_proj", | |
| "model.layers.42.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.42.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.42.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.42.mlp.gate", | |
| "model.layers.43.self_attn.compressor.kv_proj", | |
| "model.layers.43.self_attn.compressor.gate_proj", | |
| "model.layers.43.mlp.gate", | |
| "model.layers.44.self_attn.compressor.kv_proj", | |
| "model.layers.44.self_attn.compressor.gate_proj", | |
| "model.layers.44.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.44.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.44.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.44.mlp.gate", | |
| "model.layers.45.self_attn.compressor.kv_proj", | |
| "model.layers.45.self_attn.compressor.gate_proj", | |
| "model.layers.45.mlp.gate", | |
| "model.layers.46.self_attn.compressor.kv_proj", | |
| "model.layers.46.self_attn.compressor.gate_proj", | |
| "model.layers.46.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.46.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.46.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.46.mlp.gate", | |
| "model.layers.47.self_attn.compressor.kv_proj", | |
| "model.layers.47.self_attn.compressor.gate_proj", | |
| "model.layers.47.mlp.gate", | |
| "model.layers.48.self_attn.compressor.kv_proj", | |
| "model.layers.48.self_attn.compressor.gate_proj", | |
| "model.layers.48.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.48.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.48.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.48.mlp.gate", | |
| "model.layers.49.self_attn.compressor.kv_proj", | |
| "model.layers.49.self_attn.compressor.gate_proj", | |
| "model.layers.49.mlp.gate", | |
| "model.layers.50.self_attn.compressor.kv_proj", | |
| "model.layers.50.self_attn.compressor.gate_proj", | |
| "model.layers.50.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.50.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.50.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.50.mlp.gate", | |
| "model.layers.51.self_attn.compressor.kv_proj", | |
| "model.layers.51.self_attn.compressor.gate_proj", | |
| "model.layers.51.mlp.gate", | |
| "model.layers.52.self_attn.compressor.kv_proj", | |
| "model.layers.52.self_attn.compressor.gate_proj", | |
| "model.layers.52.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.52.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.52.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.52.mlp.gate", | |
| "model.layers.53.self_attn.compressor.kv_proj", | |
| "model.layers.53.self_attn.compressor.gate_proj", | |
| "model.layers.53.mlp.gate", | |
| "model.layers.54.self_attn.compressor.kv_proj", | |
| "model.layers.54.self_attn.compressor.gate_proj", | |
| "model.layers.54.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.54.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.54.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.54.mlp.gate", | |
| "model.layers.55.self_attn.compressor.kv_proj", | |
| "model.layers.55.self_attn.compressor.gate_proj", | |
| "model.layers.55.mlp.gate", | |
| "model.layers.56.self_attn.compressor.kv_proj", | |
| "model.layers.56.self_attn.compressor.gate_proj", | |
| "model.layers.56.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.56.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.56.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.56.mlp.gate", | |
| "model.layers.57.self_attn.compressor.kv_proj", | |
| "model.layers.57.self_attn.compressor.gate_proj", | |
| "model.layers.57.mlp.gate", | |
| "model.layers.58.self_attn.compressor.kv_proj", | |
| "model.layers.58.self_attn.compressor.gate_proj", | |
| "model.layers.58.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.58.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.58.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.58.mlp.gate", | |
| "model.layers.59.self_attn.compressor.kv_proj", | |
| "model.layers.59.self_attn.compressor.gate_proj", | |
| "model.layers.59.mlp.gate", | |
| "model.layers.60.self_attn.compressor.kv_proj", | |
| "model.layers.60.self_attn.compressor.gate_proj", | |
| "model.layers.60.self_attn.compressor.indexer.kv_proj", | |
| "model.layers.60.self_attn.compressor.indexer.gate_proj", | |
| "model.layers.60.self_attn.compressor.indexer.scorer.weights_proj", | |
| "model.layers.60.mlp.gate", | |
| "lm_head", | |
| "re:^mtp.*" | |
| ], | |
| "kv_cache_scheme": { | |
| "num_bits": 8, | |
| "type": "float", | |
| "dynamic": true | |
| }, | |
| "quant_method": "compressed-tensors", | |
| "quantization_status": "compressed", | |
| "sparsity_config": {}, | |
| "transform_config": {}, | |
| "version": "0.17.1.dev6+g3d7e616.d20260611" | |
| } | |
| } |