Image-Text-to-Text
Transformers
Safetensors
glm5_next
abliterated
uncensored
glm
glm-5.3
glm-5.3-flash
Mixture of Experts
exl3
tr3
quantization
reasoning
conversational
4-bit precision
Instructions to use lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] pipe(text=messages)# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated") model = AutoModelForMultimodalLM.from_pretrained("lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated", device_map="auto") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'Use Docker
docker model run hf.co/lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated
- SGLang
How to use lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated 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 "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated" \ --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": "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }'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 "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated" \ --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": "lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated", "messages": [ { "role": "user", "content": [ { "type": "text", "text": "Describe this image in one sentence." }, { "type": "image_url", "image_url": { "url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg" } } ] } ] }' - Docker Model Runner
How to use lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated with Docker Model Runner:
docker model run hf.co/lovesenko/GLM-5.3-Flash-tr3-4bpw-Abliterated
File size: 6,227 Bytes
ca44eaa | 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 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 | {
"schema": "glm53-exl3-sm120-runtime-publication.v1",
"created_at": "2026-08-27T19:08:00-04:00",
"model": "brandonmusic/GLM-5.3-Flash-EXL3-4bpw",
"hardware": "2x RTX PRO 6000 Blackwell, SM120",
"image": {
"versioned": "verdictai/glm53-flash-exl3-k4:r19-sm120-tp2-v44",
"daily_tag": "verdictai/glm53-flash-exl3-k4:daily-driver-sm120-tp2",
"oci_index": "sha256:15192e3930b4ae5558271ebe7d1a5a02da6dcc5a6c292c44e79a3fb8c883b5e1",
"amd64_manifest": "sha256:0c35421c2c773743ee74b17592d5bd32143546a2bbfe32fc8f32b97ca74167bf",
"attestation_manifest": "sha256:fa19cab01fc3304137b8dae75ba1d1162b6373d2ad177424861c63e2eaca6af9"
},
"source": {
"vllm_base_commit": "df684ff47dcbf088b41311494fd20347a702e56a",
"b12x_commit": "30aafad96b7a78064651c4a5ac177791e7bdee30",
"checkpoint_layout": "global unsliced K4 routed experts, intermediate_size=2048",
"runtime_layout": "stream-loaded TP2 rank-local K4 slabs, intermediate_size=1024",
"moe_path": "fused EXL3 Trellis/B12X"
},
"profiles": {
"daily": {
"kv_cache": "fp8_ds_mla",
"attention_backend": "FLASHINFER_MLA_SPARSE_SM120",
"dcp": 2,
"mtp_tokens": 3,
"cuda_graph": true,
"max_model_len": 435456,
"max_num_batched_tokens": 2048,
"max_num_seqs": 1,
"gpu_memory_utilization": 0.986
},
"long500k": {
"kv_cache": "nvfp4_ds_mla",
"attention_backend": "B12X_MLA_SPARSE",
"dcp": 2,
"mtp_tokens": 3,
"cuda_graph": true,
"max_model_len": 499968,
"max_num_batched_tokens": 1024,
"max_num_seqs": 1,
"gpu_memory_utilization": 0.985
}
},
"kld": {
"authoritative_regime": "v43 rootfs, TP2, DCP1, eager, no MTP, FP8 MLA KV",
"run_count": 5,
"tokens": 2048,
"prediction_positions_per_run": 2047,
"run_means": [
0.024566116963587743,
0.02484955747693601,
0.02488293126922237,
0.024016412383556018,
0.024827864888926656
],
"mean": 0.02462857659644576,
"population_stddev": 0.0003261566807242095,
"mean_top1_agreement": 0.9379579872984856,
"quality_gate": "mean KLD < 0.06",
"all_runs_pass": true,
"teacher_payload_sha256": "9f49af1b1b1a6ac88a00f5feaa89c25232597306d73d7c0cd30bb7e9c775cfb6",
"receipt_sha256": "8302e72a523189af1fe5a5e2530d45f9532c0efca8f7b4e43eb5cdc3dfd0e1e",
"first64_mean": 0.19186695,
"remaining1983_mean": 0.01923107,
"first64_fraction_of_total": 0.243613,
"nvfp4_five_run_mean": 0.06053485053836315,
"nvfp4_gate_pass": false
},
"performance": {
"nvfp4_dcp2_mtp3_steady_prefill_tok_s": {
"8192": 3819.0,
"16384": 4112.0,
"32768": 4149.0,
"65536": 4174.0,
"131072": 4145.0
},
"nvfp4_dcp2_mtp3_c1_decode_tok_s": {
"0": 98.85495572331693,
"16384": 106.5414194469794,
"32768": 101.15099399633722,
"65536": 107.15700851940537,
"131072": 112.37636035493242
},
"nvfp4_prefill_receipt_sha256": "276b30f965082c8ea82f5ad8212106a2baf3958577883a4fbd51e51a73a66f83",
"nvfp4_decode_receipt_sha256": "f5184875558e5dfe2e8f51f40a6657ea32ce5466e58b50d958225767790880d8",
"public_fp8_prefill_receipt_sha256": "c4fcefea0fe11be2197a4e23dadf35a1f5d82ba98168373f013b6537703fd2a8",
"public_fp8_decode_receipt_sha256": "e43039cf541b36cd8c66f678e7c7f6d11af2ab00250000745ad86c4f31eaf427"
},
"quality": {
"sampling": {
"temperature": 1.0,
"top_p": 0.95
},
"estonia": {
"regime": "NVFP4, DCP2, graph, MTP3",
"correct": 10,
"total": 10,
"cap_hits": 0,
"errors": 0,
"mean_generation_tok_s": 129.08494771802967,
"receipt_sha256": "2b202a67daca1eafc66758ee3019d49161fa6ea87d8455bc3080620562a4788a"
},
"lavd_low": {
"regime": "FP8, DCP2, graph, MTP3, reasoning_effort=low",
"raw_exact": 0,
"raw_near": 6,
"raw_fail": 4,
"audited_exact": 1,
"audited_near": 9,
"audited_fail": 0,
"audited_accepted": 10,
"total": 10,
"cap_hits": 0,
"errors": 0,
"mean_generation_tok_s": 160.04339599723795,
"raw_receipt_sha256": "bb8b6dad760ba36a102caf22442e6540c28aea871c6fcfb609d9ae199f65869e",
"audit_receipt_sha256": "20c595f24116e0bdf4c47c38cb9ff045387bdb637706ee96b5e14059eb8b1de8"
},
"lavd_normal_reasoning_scout": {
"regime": "FP8, DCP2, graph, MTP3, default reasoning",
"audited_exact": 2,
"audited_near": 0,
"audited_fail": 1,
"total": 3,
"cap_hits": 1,
"raw_receipt_sha256": "ea6214f4b02f44d2b2d98535a322a09aa1121d117e12b5b14a82763cb102b30c",
"audit_receipt_sha256": "360f4905751af1f478543a664e30b1decd825fd3579369d1914ec5c1b62dbb7f"
},
"hotel": {
"status": "not_run_by_owner_instruction"
}
},
"needle": {
"regime": "NVFP4, DCP2, graph, MTP3, max_model_len=499968",
"raw_passed": 17,
"raw_total": 18,
"raw_receipt_sha256": "baa77d044d553732b4c2beec66292fdb4977eafbce6e8d8ca8a72841a24c3fc1",
"targeted_retry_prompt_tokens": 498368,
"targeted_retry_depth": 0.9,
"targeted_retry_reasoning_effort": "low",
"targeted_retry_completion_tokens": 7,
"targeted_retry_exact": true,
"targeted_retry_seconds": 147.939,
"targeted_retry_receipt_sha256": "bc81617e23e2b0f337f5a7252127741301175d8bc071480ce87b9c29bc9a0ef5"
},
"kernel_tuning": {
"scheduler_1024": "block_m=32 selected for the GLM53 TP2 1024-wide expert specialization",
"scheduler_2048": "block_m=64 retained because end-to-end prefill is faster at the daily 2048-token cap",
"steady_prefill_conclusion": "approximately 4.1K tok/s; no measured 10K path on this host"
},
"live_handoff": {
"verified_at": "2026-08-27T19:13:00-04:00",
"container": "glm53-flash-exl3-k4-daily",
"health_ready": true,
"served_model": "GLM-5.3-Flash-EXL3-4bpw",
"reported_max_model_len": 435456,
"reported_kv_cache_tokens": 578698,
"cuda_graphs_captured": true,
"smoke_prompt": "What is the capital of Kentucky? Answer in one short sentence.",
"smoke_content": "The capital of Kentucky is Frankfort.",
"smoke_finish_reason": "stop",
"smoke_completion_tokens": 10
}
}
|