Instructions to use festr2/GLM-5.2-Int8Mix-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use festr2/GLM-5.2-Int8Mix-NVFP4 with Transformers:

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

pipe = pipeline("text-generation", model="festr2/GLM-5.2-Int8Mix-NVFP4")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("festr2/GLM-5.2-Int8Mix-NVFP4")
model = AutoModelForCausalLM.from_pretrained("festr2/GLM-5.2-Int8Mix-NVFP4")
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 festr2/GLM-5.2-Int8Mix-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "festr2/GLM-5.2-Int8Mix-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "festr2/GLM-5.2-Int8Mix-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/festr2/GLM-5.2-Int8Mix-NVFP4

SGLang

How to use festr2/GLM-5.2-Int8Mix-NVFP4 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 "festr2/GLM-5.2-Int8Mix-NVFP4" \
    --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": "festr2/GLM-5.2-Int8Mix-NVFP4",
		"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 "festr2/GLM-5.2-Int8Mix-NVFP4" \
        --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": "festr2/GLM-5.2-Int8Mix-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use festr2/GLM-5.2-Int8Mix-NVFP4 with Docker Model Runner:
```
docker model run hf.co/festr2/GLM-5.2-Int8Mix-NVFP4
```

GLM-5.2-Int8Mix-NVFP4 / config.json

festr2

Upload GLM-5.2 W8 plus Luke NVFP4 experts MTP config fix

7f7d834 verified 18 days ago

Raw

History Blame Contribute Delete

7.56 kB

	{
	"name_or_path": "tclf90/GLM-5.2-Int4-Int8Mix",
	"architectures": [
	"GlmMoeDsaForCausalLM"
	],
	"attention_bias": false,
	"attention_dropout": 0.0,
	"dtype": "bfloat16",
	"eos_token_id": [
	154820,
	154827,
	154829
	],
	"ep_size": 1,
	"first_k_dense_replace": 3,
	"head_dim": 192,
	"hidden_act": "silu",
	"hidden_size": 6144,
	"index_head_dim": 128,
	"index_n_heads": 32,
	"index_share_for_mtp_iteration": true,
	"index_skip_topk_offset": 3,
	"index_topk": 2048,
	"index_topk_freq": 4,
	"index_topk_pattern": null,
	"indexer_rope_interleave": true,
	"indexer_types": [
	"full",
	"full",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared",
	"full",
	"shared",
	"shared",
	"shared"
	],
	"initializer_range": 0.02,
	"intermediate_size": 12288,
	"kv_lora_rank": 512,
	"max_position_embeddings": 1048576,
	"mlp_layer_types": [
	"dense",
	"dense",
	"dense",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse",
	"sparse"
	],
	"model_type": "glm_moe_dsa",
	"moe_intermediate_size": 2048,
	"moe_layer_freq": 1,
	"n_group": 1,
	"n_routed_experts": 256,
	"n_shared_experts": 1,
	"norm_topk_prob": true,
	"num_attention_heads": 64,
	"num_experts_per_tok": 8,
	"num_hidden_layers": 78,
	"num_key_value_heads": 64,
	"num_nextn_predict_layers": 1,
	"pad_token_id": 154820,
	"pretraining_tp": 1,
	"q_lora_rank": 2048,
	"qk_head_dim": 256,
	"qk_nope_head_dim": 192,
	"qk_rope_head_dim": 64,
	"rms_norm_eps": 1e-05,
	"rope_interleave": true,
	"rope_parameters": {
	"rope_theta": 8000000,
	"rope_type": "default"
	},
	"routed_scaling_factor": 2.5,
	"scoring_func": "sigmoid",
	"tie_word_embeddings": false,
	"topk_group": 1,
	"topk_method": "noaux_tc",
	"transformers_version": "5.12.0",
	"use_cache": true,
	"v_head_dim": 256,
	"vocab_size": 154880,
	"quantization_config": {
	"quant_method": "compressed-tensors",
	"format": "nvfp4-pack-quantized",
	"ignore": [
	"re:model[.]layers[.]0[.].*",
	"re:model[.]layers[.][1-9][0-9][.](?:mtp_block[.])?mlp[.]gate(?:$\|[.].)",
	"re:model[.]layers[.][1-9][0-9][.](?:mtp_block[.])?self_attn[.]indexer(?:$\|[.].)",
	"re:model[.]layers[.][1-9][0-9][.](?:mtp_block[.])?self_attn[.]indexers_proj(?:$\|[.].)",
	"re:model[.]layers[.][1-9][0-9][.](?:eh_proj\|enorm\|hnorm)[.].",
	"re:model[.]layers[.][1-9][0-9][.]shared_head[.]norm[.].",
	"re:model[.]layers[.][1-9][0-9][.]shared_head[.]head(?:$\|[.].)"
	],
	"config_groups": {
	"w4a16_experts": {
	"targets": [
	"re:model[.]layers[.](?:3\|4\|5\|6\|7\|8\|9\|10\|11\|12\|13\|14\|15\|16\|17\|18\|19\|20\|21\|22\|23\|24\|25\|26\|27\|28\|29\|30\|31\|32\|33\|34\|35\|36\|37\|38\|39\|40\|41\|42\|43\|44\|45\|46\|47\|48\|49\|50\|51\|52\|53\|54\|55\|56\|57\|58\|59\|60\|61\|62\|63\|64\|65\|66\|67\|68\|69\|70\|71\|72\|73\|74\|75\|76\|77)[.]mlp[.]experts[.][0-9]+[.](?:gate_proj\|up_proj\|down_proj)$"
	],
	"weights": {
	"num_bits": 4,
	"type": "float",
	"symmetric": true,
	"strategy": "tensor_group",
	"group_size": 16,
	"dynamic": false
	},
	"format": "nvfp4-pack-quantized"
	},
	"w8a16_linears": {
	"targets": [
	"re:model[.]layers[.](?:1\|2\|3\|4\|5\|6\|7\|8\|9\|10\|11\|12\|13\|14\|15\|16\|17\|18\|19\|20\|21\|22\|23\|24\|25\|26\|27\|28\|29\|30\|31\|32\|33\|34\|35\|36\|37\|38\|39\|40\|41\|42\|43\|44\|45\|46\|47\|48\|49\|50\|51\|52\|53\|54\|55\|56\|57\|58\|59\|60\|61\|62\|63\|64\|65\|66\|67\|68\|69\|70\|71\|72\|73\|74\|75\|76\|77)[.](?:self_attn[.](?:fused_qkv_a_proj\|fused_qkv_a_proj_with_mqa\|q_a_proj\|q_b_proj\|kv_a_proj_with_mqa\|kv_b_proj\|o_proj)\|mlp[.](?:gate_up_proj\|gate_proj\|up_proj\|down_proj\|shared_experts[.](?:gate_up_proj\|gate_proj\|up_proj\|down_proj)))$"
	],
	"weights": {
	"num_bits": 8,
	"type": "int",
	"symmetric": true,
	"strategy": "group",
	"group_size": 128,
	"dynamic": false
	},
	"format": "pack-quantized"
	},
	"w8a16_mtp_channel": {
	"targets": [
	"re:model[.]layers[.](?:78)[.](?:mtp_block[.])?(?:self_attn[.](?:fused_qkv_a_proj\|fused_qkv_a_proj_with_mqa\|q_a_proj\|q_b_proj\|kv_a_proj_with_mqa\|kv_b_proj\|o_proj)\|mlp[.](?:experts[.][0-9]+[.](?:gate_proj\|up_proj\|down_proj)\|gate_up_proj\|gate_proj\|up_proj\|down_proj\|shared_experts[.](?:gate_up_proj\|gate_proj\|up_proj\|down_proj)))$"
	],
	"weights": {
	"num_bits": 8,
	"type": "int",
	"symmetric": true,
	"strategy": "channel",
	"group_size": -1,
	"dynamic": false
	},
	"format": "pack-quantized"
	}
	},
	"packed_modules_mapping": {
	"fused_qkv_a_proj_with_mqa": [
	"q_a_proj",
	"kv_a_proj_with_mqa"
	],
	"gate_up_proj": [
	"gate_proj",
	"up_proj"
	]
	},
	"producer": {
	"name": "codex-quanttrio-w8-plus-luke-nvfp4-experts",
	"quanttrio_base": "/root/.cache/huggingface/hub/models--QuantTrio--GLM-5.2-Int4-Int8Mix/snapshots/8677dbb545f2cb9825fcb76dff122963cf920065",
	"luke_nvfp4_source": "/root/.cache/huggingface/hub/models--lukealonso--GLM-5.2-NVFP4/snapshots/8a1f4a13204acf2b7ac840375efaed64c231c522",
	"policy": "Keep QuantTrio compressed-tensors W8A16 linears and layer78 W8A16 MTP; replace only QuantTrio W4A16 INT4 MoE expert tensors in layers 3..77 with Luke NVFP4 expert tensors using compressed-tensors NVFP4 W4A16 names.",
	"replaced_layers": [
	3,
	77
	],
	"replaced_expert_weight_count": 57600
	}
	},
	"_codex_notes": {
	"mtpfix": "Add runtime fused_qkv_a_proj to W8A16 quant targets so GLM-5.2 layer78 MTP fused-QKV registers weight_packed params."
	}
	}