GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
amd
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
amd
Model Tree
Qwen/Qwen3.6-35B-A3B
Input Modalities
Output Modalities
Supported Functionality
This is a quantized version of Qwen3.6-35B-A3B created by AMD using LLM Compressor (compressed-tensors) for ZenDNN-optimized CPU inference.
The model was quantized from Qwen3.6-35B-A3B using LLM Compressor via the GPTQ algorithm. This reduces the model weights from 67.0 GiB to 18.1 GiB on disk (~73% reduction).
compressed-tensors, num_bits=4, type=int, symmetric=true, group_size=128group_size=128, actorder=static), stored as pack-quantizedmodel.visual), the MoE router (mlp.gate), , , , and the layer norms. The 256 routed experts, the shared expert, and the linear-attention projections are all quantized, which is why the reduction is close to the 75% ideal for INT4.Three details in the recipe are specific to this model. The model is loaded with AutoModelForImageTextToText rather than AutoModelForCausalLM, so vision_config and the Qwen3_5MoeForConditionalGeneration metadata survive into the saved config for vLLM 0.26+; without it the config is demoted to the text-only variant and vLLM rejects the checkpoint. shared_expert_gate is skipped because it has out_features=1 and vLLM expects a dense .weight there rather than a packed one. And the router is skipped because it is a tiny Linear whose logits decide expert assignment, where a mis-routed token costs far more accuracy than 4-bit expert weights do.
python
import jsonimport os import torchfrom datasets import load_datasetfrom transformers import AutoModelForImageTextToText, AutoTokenizer from llmcompressor import oneshotfrom llmcompressor.modifiers.quantization.gptq import GPTQModifier MODEL_ID = "Qwen/Qwen3.6-35B-A3B"OUTPUT_DIR = "./Qwen3.6-35B-A3B-w4a16-llmcompressor-v0.12.0"CALIB_SIZE = 128MAX_SEQ_LENGTH = 2048 IGNORE = [ "lm_head", r"re:.*\.router$", r"re:.*\.router\..*", r"re:.*\.gate$", r"re:.*\.mlp\.gate$", "shared_expert_gate", r"re:.*\.shared_expert_gate(\.|$)", r"re:.*\.visual(\.|$)", r"re:.*\.visual\..*",] # Step 1: Load the BF16 model with the multimodal wrapper so vision_config and# the ForConditionalGeneration architecture are preserved on save.model = AutoModelForImageTextToText.from_pretrained( MODEL_ID, dtype=torch.bfloat16, device_map="cpu", trust_remote_code=True,)reference_config = model.configtokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True) # Step 2: Load calibration data. GPTQ is data-driven: it needs real activations# to build the per-layer Hessians used to compensate the rounding error.ds = load_dataset("HuggingFaceH4/ultrachat_200k", split=f"train_sft[:{CALIB_SIZE}]")ds = ds.map( lambda ex: {"text": "\n".join(m["content"] for m in ex["messages"] if m.get("content"))}, remove_columns=ds.column_names,)if not getattr(tokenizer, "pad_token", None): tokenizer.pad_token = tokenizer.eos_tokencalib_ds = ds.map( lambda ex: tokenizer( ex["text"], truncation=True, max_length=MAX_SEQ_LENGTH, add_special_tokens=False ), remove_columns=["text"],) # Step 3: Define the W4A16 GPTQ recipe and one-shot quantize. Passing# processor=MODEL_ID saves the processor configs alongside the model.recipe = GPTQModifier(targets="Linear", scheme="W4A16", ignore=IGNORE)oneshot( model=model, dataset=calib_ds, recipe=recipe, max_seq_length=MAX_SEQ_LENGTH, processor=MODEL_ID, output_dir=OUTPUT_DIR, save_compressed=True, trust_remote_code_model=True,) # Step 4: Patch config.json so architectures/vision_config stay intact for vLLM.config_path = os.path.join(OUTPUT_DIR, "config.json")with open(config_path, encoding="utf-8") as f: saved = json.load(f)full = reference_config.to_dict()for key in ( "architectures", "model_type", "text_config", "vision_config", "vision_start_token_id", "vision_end_token_id", "image_token_id", "video_token_id",): if full.get(key) is not None: saved[key] = full[key]with open(config_path, "w", encoding="utf-8") as f: json.dump(saved, f, indent=2) f.write("\n") # Smoke testinputs = tokenizer("What are we having for dinner?", return_tensors="pt")output = model.generate(**inputs, max_new_tokens=30)print(tokenizer.decode(output[0], skip_special_tokens=True))Run end to end with the standalone script:
bash
numactl --physcpubind=0-95 python quant-qwen3.6-w4a16.pypython
from vllm import LLM, SamplingParams model = LLM( model="amd/Qwen3.6-35B-A3B-w4a16-llmcompressor-v0.12.0", dtype="bfloat16",) sampling_params = SamplingParams(temperature=0.7, max_tokens=256)outputs = model.generate(["Hello, how are you?"], sampling_params)print(outputs[0].outputs[0].text)text
torch==2.13.0.0zentorch==2.13.0.0vllm==0.28.0llmcompressor==0.12.0For optimal performance, set LD_PRELOAD with libomp.so (LLVM OpenMP) or libiomp5.so (Intel OpenMP):
bash
# Using LLVM OpenMP (llvmopenmp)export LD_PRELOAD=$(find /path/to/env -name "libomp.so" | head -1) # Or using Intel OpenMP (libiomp)export LD_PRELOAD=$(find /path/to/env -name "libiomp5.so" | head -1)Note: Set
LD_PRELOADbefore launching vLLM or any inference script.
The model was evaluated against the BF16 (unquantized) baseline on standard benchmarks using lm-evaluation-harness with the vLLM engine.
| Benchmark | BF16 Baseline | W4A16 (this model) | Recovery |
|---|---|---|---|
| GSM8K (5-shot) | 0.9651 | 0.9613 | 99.61% |
bash
lm_eval \ --model vllm \ --model_args pretrained=amd/Qwen3.6-35B-A3B-w4a16-llmcompressor-v0.12.0,tokenizer=Qwen/Qwen3.6-35B-A3B,dtype=bfloat16,max_model_len=4096,language_model_only=True,enable_thinking=False \ --tasks gsm8k \ --batch_size auto \ --trust_remote_code \ --num_fewshot 5 \ --apply_chat_template \ --log_samples \ --gen_kwargs "max_gen_toks=2048" \ --output_path .language_model_only=True.enable_thinking=False (and a generous max_gen_toks) so answer extraction stays comparable to the BF16 baseline.This model is distributed under the same license as the source model. See the LICENSE file for details.
Modifications copyright (c) 2026 Advanced Micro Devices, Inc. All rights reserved.
shared_expert_gatelm_headembed_tokens