MiniCPM-V-4.7-35B-A3B ยท FP8 (dynamic)

8-bit quantization of openbmb/MiniCPM-V-4.7-35B-A3B for vLLM serving. Quantized by axteam.

Revision 2026-11-02 (config fix): the earlier config.json excluded the vision tower from quantization with the pattern re:model\.vision_tower\..*, which does not match the names vLLM uses for these modules (vpm.*, vit_merger.*). vLLM 0.32 therefore built the vision layers as FP8 and loaded the BF16 vision weights into them, so image input gave wrong answers; text input was not affected. Only config.json changed (vLLM's names were added to the ignore list as re:vpm\..* and re:vit_merger\..*, and the existing patterns re:model\.visual\..*, re:model\.vision_tower\..* and re:model\.merger\..* were widened to re:(model\.)?visual\..*, re:(model\.)?vision_tower\..* and re:(model\.)?merger\..*); the weights are the same. Checked with vLLM 0.32.0 on 2026-11-02: the vision weights now load as BF16, and a synthetic test image (red circle, blue square, the caption "QWEN TEST 2026") is described correctly (with vLLM 0.32.0 + transformers 5.21 the image preprocessing fails, see Requirements; for this check only, the image processor's missing downsample_mode argument was filled in with its default). If you downloaded this repo before 2026-11-01 23:41:44 UTC (revision a1f77bb), download config.json again.

Base model openbmb/MiniCPM-V-4.7-35B-A3B
Method FP8 (dynamic)
Format compressed-tensors float-quantized (scheme FP8_DYNAMIC)
Weights / activations W8A8 โ€” FP8 E4M3 weights (per-channel), dynamic per-token FP8 activations
Architecture MoE (Qwen3.5-MoE backbone) ยท vision ยท multimodal
Size on disk ~36.9 GB
Calibration None (data-free)
License Apache-2.0

Quantization details

  • Scheme โ€” W8A8 โ€” FP8 E4M3 weights (per-channel), dynamic per-token FP8 activations, stored as compressed-tensors float-quantized (scheme FP8_DYNAMIC).
  • Kept in BF16 โ€” the vision tower & vision-text merger, token embeddings, the LM head, all RMSNorms, the MoE routers (mlp.gate, shared_expert_gate) โ€” about 5% of all parameters in this build. Quantized: the attention (incl. linear-attention) and MoE/MLP projections.
  • Post-training quantization (no retraining). Tokenizer / chat_template identical to the original.

Benchmarks (measured)

Every column was measured with the same harness and settings โ€” lm-evaluation-harness on vLLM, greedy, completion-style prompts without a chat template (0- or 5-shot, as listed in each row). MMLU here is lm-eval's generative variant (mmlu_generative: the model writes the answer letter and only an exact match on the first line counts), so it is not directly comparable with MMLU scores measured another way (e.g. log-likelihood scoring). Percentages are recovery vs. the BF16 original.

Benchmark BF16 (original) FP8-dynamic (this repo) AWQ-8bit NVFP4A16 NVFP4 AWQ-4bit
MMLU (5-shot, generative) 85.65 85.61 (100.0%) 85.71 (100.1%) 85.76 (100.1%) 85.41 (99.7%) 85.31 (99.6%)
GSM8K (5-shot, strict) 90.14 89.69 (99.5%) 89.23 (99.0%) 90.75 (100.7%) 90.22 (100.1%) 90.52 (100.4%)
ARC-Challenge (0-shot) 60.92 61.01 (100.1%) 61.18 (100.4%) 60.84 (99.9%) 61.09 (100.3%) 62.03 (101.8%)
Winogrande (5-shot) 80.11 79.87 (99.7%) 78.93 (98.5%) 79.32 (99.0%) 77.11 (96.3%) 79.56 (99.3%)
KMMLU (5-shot, direct) 64.95 64.77 (99.7%) 64.78 (99.7%) 64.65 (99.5%) 64.38 (99.1%) 64.67 (99.6%)
HAE-RAE (0-shot) 73.14 73.69 (100.8%) 73.42 (100.4%) 73.51 (100.5%) 74.06 (101.3%) 72.69 (99.4%)
Average 75.82 75.77 (99.9%) 75.54 (99.6%) 75.80 (100.0%) 75.38 (99.4%) 75.80 (100.0%)

Evaluation (measured)

Fidelity to the BF16 original on an internal mixed English/Korean set (100 prompts: the BF16 original's greedy answers, scored under each build with the same prompts and scoring code). This build was measured with vLLM 0.32 on 2026-11-01; the BF16 reference column was measured on 2026-11-01.

Metric This model BF16 original
Perplexity (lower = closer) 1.2269 1.2063 (+1.7%)
Top-1 next-token agreement 96.0% 98.1%

These measure fidelity to the base model, not absolute task scores.

Requirements

Runtime vLLM โ€” the architecture is in vLLM โ‰ฅ 0.32.0; tested here with vLLM 0.32.0. vLLM 0.32 needs transformers 5.18.x or 5.19.x for this model (5.17 does not know minicpmv4_7); our runs used transformers 5.21.0 installed with --no-deps, text-only โ€” image input fails in that setup (vLLM's MiniCPM-V image processor vs transformers 5.21); only the 2026-11-02 image check in the revision note above ran images, with the processor's missing downsample_mode argument filled in for the test.
vLLM install vLLM 0.32 itself needs Python 3.11โ€“3.14, torch 2.13 (CUDA 13.0 build, so an NVIDIA driver with CUDA 13 support), transformers โ‰ฅ5.16.1,<5.20 and compressed-tensors 0.17.0 (both pinned by vLLM; 0.17.0 reads this checkpoint, which was saved with 0.19.0 โ€” do not upgrade it separately).
GPU NVIDIA Turing (sm75, e.g. T4) or newer. FP8 math runs on Ada / Hopper / Blackwell (sm89+); on Turing and Ampere vLLM switches to weight-only FP8 automatically. V100 (sm70) is not supported. (Kernel choice per GPU from vLLM 0.32's selection code; only B200 was run here.)
Memory Weights: 36.9 GB. Add KV cache (grows with context length ร— concurrent requests) and a few GB for activations / CUDA graphs; use --tensor-parallel-size N when the weights do not fit one GPU.
Transformers (without vLLM) transformers>=5.18.0 (FP8 checked).
trust_remote_code not needed

Usage (vLLM)

Compatibility (checked 2026-10-31): with vLLM 0.32.0 and Transformers 5.17.0 the server did not start โ€” Transformers does not recognise model type minicpmv4_7 and this repo ships no custom modeling code. Use a Transformers / vLLM release that supports MiniCPM-V 4.7.

Start an OpenAI-compatible server:

vllm serve axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic --max-model-len 32768

Query it:

curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json" -d '{
  "model": "axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic",
  "messages": [{"role": "user", "content": "๋Œ€ํ•œ๋ฏผ๊ตญ์˜ ์ˆ˜๋„๋Š” ์–ด๋””์ธ๊ฐ€์š”?"}],
  "temperature": 1.0, "top_p": 0.95, "top_k": 20, "max_tokens": 512
}'

Offline inference in Python:

from vllm import LLM, SamplingParams

llm = LLM(model="axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic", max_model_len=32768)
params = SamplingParams(temperature=1.0, top_p=0.95, top_k=20, max_tokens=1024)
out = llm.chat([{"role": "user", "content": "Explain model quantization in two sentences."}], params)
print(out[0].outputs[0].text)

# image input (URL or data: URI)
messages = [{"role": "user", "content": [
    {"type": "image_url", "image_url": {"url": "https://example.com/your-image.png"}},
    {"type": "text", "text": "Describe this image."},
]}]
out = llm.chat(messages, params)
print(out[0].outputs[0].text)

Recommended sampling (from the original generation_config.json): temperature=1.0, top_p=0.95, top_k=20. Thinking mode is off by default in this model's chat template; to turn it on pass chat_template_kwargs={"enable_thinking": True} (Python llm.chat(..., chat_template_kwargs=...), or "chat_template_kwargs": {"enable_thinking": true} in the request body). Text-only serving: add --limit-mm-per-prompt '{"image": 0, "video": 0}' to skip loading the vision encoder's memory budget.

Reducing KV-cache memory (optional, serve-time)

The KV cache is kept in BF16 by default. KV-cache precision is a runtime choice, independent of the weight quantization above. This is a hybrid model: only its 10 full-attention layers keep a KV cache, while its 30 linear-attention (Gated DeltaNet) layers keep a fixed-size state per request, which --kv-cache-dtype does not change. FP8 halves the full-attention KV cache, so the saving grows with context length and is small for many short requests:

vllm serve axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic --kv-cache-dtype fp8

License

Quantized derivative of openbmb/MiniCPM-V-4.7-35B-A3B (Apache-2.0). Original weights & tokenizer belong to OpenBMB.


ํ•œ๊ตญ์–ด ์š”์•ฝ

openbmb/MiniCPM-V-4.7-35B-A3B์˜ FP8 (W8A8) ์–‘์žํ™” ๋นŒ๋“œ ยท ์ œ์ž‘ axteam

๊ฐœ์š”

ํ•ญ๋ชฉ ๋‚ด์šฉ
์›๋ณธ ๋ชจ๋ธ openbmb/MiniCPM-V-4.7-35B-A3B
์–‘์žํ™” ๋ฐฉ์‹ FP8 ๋™์  ์–‘์žํ™” (W8A8)
์ €์žฅ ํ˜•์‹ compressed-tensors float-quantized (scheme FP8_DYNAMIC)
๋ชจ๋ธ ๊ตฌ์กฐ MoE (Qwen3.5-MoE ๋ฐฑ๋ณธ) ยท ๋ฉ€ํ‹ฐ๋ชจ๋‹ฌ(๋น„์ „)
์šฉ๋Ÿ‰ ์•ฝ 36.9 GB
๋ณด์ • ๋ฐ์ดํ„ฐ ์—†์Œ (๊ฐ€์ค‘์น˜ ์Šค์ผ€์ผ์€ ๊ฐ€์ค‘์น˜๋งŒ์œผ๋กœ ๊ณ„์‚ฐ, ํ™œ์„ฑํ™” ์Šค์ผ€์ผ์€ ์‹คํ–‰ ์‹œ ๊ณ„์‚ฐ)

์–‘์žํ™” ์ƒ์„ธ

  • ๊ฐ€์ค‘์น˜: FP8(E4M3), ์ถœ๋ ฅ ์ฑ„๋„๋ณ„ ์Šค์ผ€์ผ
  • ํ™œ์„ฑํ™”: ์‹คํ–‰ ์‹œ ํ† ํฐ๋ณ„ ๋™์  FP8
  • ์–‘์žํ™” ์ ˆ์ฐจ: PTQ(ํ•™์Šต ํ›„ ์–‘์žํ™”, ์žฌํ•™์Šต ์—†์Œ), ๊ฐ€์ค‘์น˜ ์Šค์ผ€์ผ์€ ๋ฏธ๋ฆฌ ๊ณ„์‚ฐํ•˜๊ณ  ํ™œ์„ฑํ™” ์Šค์ผ€์ผ์€ ์‹คํ–‰ ์‹œ ํ† ํฐ๋งˆ๋‹ค ๊ณ„์‚ฐ
  • BF16 ์œ ์ง€: ๋น„์ „ ํƒ€์›Œ, ๋น„์ „-ํ…์ŠคํŠธ ํˆฌ์˜์ธต(merger), ํ† ํฐ ์ž„๋ฒ ๋”ฉ, LM ํ—ค๋“œ, ๋ชจ๋“  RMSNorm, MoE ๋ผ์šฐํ„ฐ(mlp.gate)์™€ ๊ณต์œ  ์ „๋ฌธ๊ฐ€ ๊ฒŒ์ดํŠธ(shared_expert_gate) โ€” ์ „์ฒด ํŒŒ๋ผ๋ฏธํ„ฐ์˜ ์•ฝ 5%
  • ์–‘์žํ™” ๋Œ€์ƒ: ์–ดํ…์…˜(์„ ํ˜• ์–ดํ…์…˜ ํฌํ•จ)๊ณผ MoE/MLP์˜ ์„ ํ˜•์ธต(projection)
  • ์›๋ณธ ํŒŒ์ผ: ํ† ํฌ๋‚˜์ด์ €ยทchat template์€ ์›๋ณธ๊ณผ ๋™์ผ

ํ’ˆ์งˆ (์‹ค์ธก)

  • ์›๋ณธ ๋Œ€๋น„ ์ถฉ์‹ค๋„: PPL 1.2269 (์›๋ณธ 1.2063, +1.7%, ๋‚ฎ์„์ˆ˜๋ก ์›๋ณธ์— ๊ฐ€๊นŒ์›€), ๋‹ค์Œ ํ† ํฐ top-1 ์ผ์น˜์œจ 96.0% (์›๋ณธ ์ž์ฒด ๊ธฐ์ค€๊ฐ’ 98.1%) โ€” ๋‚ด๋ถ€ ์˜์–ดยทํ•œ๊ตญ์–ด ํ˜ผํ•ฉ ํ”„๋กฌํ”„ํŠธ 100๊ฐœ, ์›๋ณธ(BF16)์˜ greedy ๋‹ต๋ณ€์„ ๊ธฐ์ค€ ํ…์ŠคํŠธ๋กœ ์ธก์ •
  • ๋ฒค์น˜๋งˆํฌ ํ‰๊ท : 75.77 (์›๋ณธ 75.82, 99.9%) โ€” MMLU(์ƒ์„ฑํ˜•)ยทGSM8KยทARC-ChallengeยทWinograndeยทKMMLUยทHAE-RAE 6์ข… ํ‰๊ท , lm-evaluation-harness + vLLM, chat template ์—†๋Š” ์™„์„ฑํ˜• ํ”„๋กฌํ”„ํŠธ(๊ณผ์ œ๋ณ„ 0ยท5-shot, ํ‘œ ์ฐธ๊ณ )ยทgreedy, ๋ชจ๋“  ๋นŒ๋“œ๋ฅผ ๊ฐ™์€ ์„ค์ •์œผ๋กœ ์ธก์ •
  • ํ•œ๊ตญ์–ด ๋ฒค์น˜๋งˆํฌ: KMMLU 64.77 (์›๋ณธ 64.95, 99.7%) ยท HAE-RAE 73.69 (์›๋ณธ 73.14, 100.8%)
  • MMLU ๋ฐฉ์‹: lm-eval ์ƒ์„ฑํ˜• ๋ณ€ํ˜•(mmlu_generative) โ€” ๋ชจ๋ธ์ด ๋‹ต ๊ธ€์ž๋ฅผ ์ƒ์„ฑํ•˜๊ณ  ์ฒซ ์ค„์ด ์ •ํ™•ํžˆ ์ผ์น˜ํ•ด์•ผ ์ •๋‹ต์œผ๋กœ ์ฑ„์ ํ•˜๋ฏ€๋กœ, ๋‹ค๋ฅธ ๋ฐฉ์‹(์˜ˆ: log-likelihood ์ฑ„์ )์œผ๋กœ ์žฐ MMLU ์ ์ˆ˜์™€ ์ง์ ‘ ๋น„๊ต ๋ถˆ๊ฐ€

์š”๊ตฌ์‚ฌํ•ญ

  • ์‹คํ–‰ ์—”์ง„: vLLM 0.32.0 ์ด์ƒ(์ด ๊ตฌ์กฐ๋ฅผ ์ง€์›ํ•˜๋Š” ์ฒซ ๋ฒ„์ „), ์ด ๋นŒ๋“œ๋Š” vLLM 0.32.0์—์„œ ์‹คํ–‰ ํ™•์ธ
  • Transformers ๋ฒ„์ „: vLLM 0.32์™€ ํ•จ๊ป˜ ์“ฐ๋ ค๋ฉด transformers 5.18.x ๋˜๋Š” 5.19.x ํ•„์š” (5.17์€ minicpmv4_7 ๋ฏธ์ง€์›), ์‹คํ–‰ ํ™•์ธ์€ transformers 5.21.0์„ --no-deps๋กœ ์„ค์น˜ํ•ด ํ…์ŠคํŠธ ์ž…๋ ฅ์œผ๋กœ ์ง„ํ–‰ โ€” ์ด ์กฐํ•ฉ์—์„œ๋Š” ์ด๋ฏธ์ง€ ์ž…๋ ฅ ์‹คํŒจ (vLLM์˜ MiniCPM-V ์ด๋ฏธ์ง€ ํ”„๋กœ์„ธ์„œ์™€ transformers 5.21 ๋น„ํ˜ธํ™˜), ์ด๋ฏธ์ง€๋Š” ์œ„ ์ˆ˜์ • ์ด๋ ฅ์˜ 2026-11-02 ์ ๊ฒ€๋งŒ ์ฒ˜๋ฆฌ๊ธฐ์— ๋น ์ง„ downsample_mode ์ธ์ž๋ฅผ ์ฑ„์›Œ ์‹คํ–‰
  • vLLM 0.32 ์„ค์น˜ ์กฐ๊ฑด: Python 3.11~3.14, torch 2.13(CUDA 13.0 ๋นŒ๋“œ, CUDA 13 ์ง€์› ๋“œ๋ผ์ด๋ฒ„ ํ•„์š”), transformers 5.16.1 ์ด์ƒ 5.20 ๋ฏธ๋งŒ, compressed-tensors 0.17.0 โ€” transformersยทcompressed-tensors๋Š” vLLM์ด ๋ฒ„์ „์„ ๊ณ ์ •ํ•˜๋ฏ€๋กœ compressed-tensors๋ฅผ ๋”ฐ๋กœ ์—…๊ทธ๋ ˆ์ด๋“œํ•˜์ง€ ๋ง ๊ฒƒ (์ด ์ฒดํฌํฌ์ธํŠธ๋Š” 0.19.0์œผ๋กœ ์ €์žฅ๋์ง€๋งŒ 0.17.0์—์„œ๋„ ๋กœ๋“œ๋จ)
  • GPU: NVIDIA Turing(sm75, ์˜ˆ: T4) ์ด์ƒ, V100(sm70) ๋ฏธ์ง€์› โ€” FP8 ์—ฐ์‚ฐ์€ AdaยทHopperยทBlackwell(sm89 ์ด์ƒ)์—์„œ ์ˆ˜ํ–‰, TuringยทAmpere๋Š” vLLM์ด ๊ฐ€์ค‘์น˜ ์ „์šฉ FP8๋กœ ์ž๋™ ์ „ํ™˜ (GPU๋ณ„ ์ง€์› ์—ฌ๋ถ€๋Š” vLLM 0.32 ์ปค๋„ ์„ ํƒ ์ฝ”๋“œ ๊ธฐ์ค€, ์‹ค์ œ ์‹คํ–‰์€ B200์—์„œ๋งŒ ํ™•์ธ)
  • ๋ฉ”๋ชจ๋ฆฌ: ๊ฐ€์ค‘์น˜ 36.9 GB + KV ์บ์‹œ(์ปจํ…์ŠคํŠธ ๊ธธ์ด ร— ๋™์‹œ ์š”์ฒญ ์ˆ˜์— ๋น„๋ก€) + ํ™œ์„ฑํ™”ยทCUDA graph์šฉ ์ˆ˜ GB โ€” GPU ํ•œ ์žฅ์— ๋‹ค ๋“ค์–ด๊ฐ€์ง€ ์•Š์œผ๋ฉด --tensor-parallel-size N์œผ๋กœ ๋ถ„์‚ฐ
  • Transformers๋กœ ์ง์ ‘ ์‚ฌ์šฉ: transformers 5.18.0 ์ด์ƒ (FP8 ๋นŒ๋“œ๋กœ ํ™•์ธ)
  • trust_remote_code: ๋ถˆํ•„์š”

์‚ฌ์šฉ๋ฒ•

  • ์„œ๋ฒ„ ์‹คํ–‰: vllm serve axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic --max-model-len 32768 โ€” OpenAI ํ˜ธํ™˜ API
  • Python: LLM(model="axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic", max_model_len=32768).chat(...) โ€” ์ „์ฒด ์˜ˆ์‹œ๋Š” ์œ„ ์˜์–ด ๋ณธ๋ฌธ 'Usage (vLLM)' ์ ˆ ์ฐธ๊ณ 
  • ๊ถŒ์žฅ ์ƒ˜ํ”Œ๋ง: temperature=1.0, top_p=0.95, top_k=20 (์›๋ณธ generation_config.json ๊ธฐ์ค€)
  • thinking ๋ชจ๋“œ: ๊ธฐ๋ณธ ๊บผ์ง โ€” ์ผœ๋ ค๋ฉด chat_template_kwargs={"enable_thinking": True} ์ „๋‹ฌ(Python์€ llm.chat() ์ธ์ž, HTTP๋Š” ์š”์ฒญ ๋ณธ๋ฌธ์˜ "chat_template_kwargs": {"enable_thinking": true})
  • ์ด๋ฏธ์ง€ ์ž…๋ ฅ: ๋ฉ”์‹œ์ง€์— image_url ํ•ญ๋ชฉ ์‚ฌ์šฉ
  • ํ…์ŠคํŠธ๋งŒ ์‚ฌ์šฉํ•  ๋•Œ: --limit-mm-per-prompt '{"image": 0, "video": 0}' โ€” ๋น„์ „ ํƒ€์›Œ์šฉ ๋ฉ”๋ชจ๋ฆฌ ์˜ˆ์•ฝ์„ ์ƒ๋žตํ•ด ๋ฉ”๋ชจ๋ฆฌ ์ ˆ์•ฝ
  • KV ์บ์‹œ ์ ˆ๊ฐ(์„ ํƒ): vllm serve์— --kv-cache-dtype fp8 ์ถ”๊ฐ€ โ€” ํ•˜์ด๋ธŒ๋ฆฌ๋“œ ๊ตฌ์กฐ๋ผ KV ์บ์‹œ๋Š” ํ’€ ์–ดํ…์…˜ 10๊ฐœ ์ธต์—๋งŒ ์žˆ๊ณ , ์„ ํ˜• ์–ดํ…์…˜(Gated DeltaNet) 30๊ฐœ ์ธต์˜ ์š”์ฒญ๋ณ„ ๊ณ ์ • ํฌ๊ธฐ ์ƒํƒœ๋Š” ์ด ์˜ต์…˜์œผ๋กœ ๋ฐ”๋€Œ์ง€ ์•Š์Œ โ†’ ํ’€ ์–ดํ…์…˜ KV ์บ์‹œ๋งŒ ์ ˆ๋ฐ˜์œผ๋กœ ์ค„์–ด ๊ธด ์ปจํ…์ŠคํŠธ์—์„œ ํšจ๊ณผ๊ฐ€ ํฌ๊ณ  ์งง์€ ์š”์ฒญ์ด ๋งŽ์„ ๋•Œ๋Š” ์ž‘์Œ (๊ฐ€์ค‘์น˜ ์–‘์žํ™”์™€ ๋ณ„๊ฐœ์ธ ์‹คํ–‰ ์˜ต์…˜)

์ฐธ๊ณ 

  • ํ˜ธํ™˜์„ฑ (2026-10-31 ํ™•์ธ): vLLM 0.32.0 + transformers 5.17.0์€ minicpmv4_7์„ ์ธ์‹ํ•˜์ง€ ๋ชปํ•ด ์„œ๋ฒ„ ์‹คํ–‰ ์‹คํŒจ โ€” transformers 5.18.xยท5.19.x ํ•„์š” (์œ„ ์š”๊ตฌ์‚ฌํ•ญ ์ฐธ๊ณ )
  • ์ˆ˜์ • ์ด๋ ฅ (2026-11-02, ์„ค์ • ์ˆ˜์ •): ์ด์ „ config.json์€ ๋น„์ „ ํƒ€์›Œ๋ฅผ ์–‘์žํ™”์—์„œ ๋นผ๋Š” ๊ทœ์น™(re:model\.vision_tower\..*)์ด vLLM์ด ์“ฐ๋Š” ๋ชจ๋“ˆ ์ด๋ฆ„(vpm.*, vit_merger.*)๊ณผ ๋งž์ง€ ์•Š์•„, vLLM 0.32๊ฐ€ ๋น„์ „ ์ธต์„ FP8๋กœ ๋งŒ๋“ค๊ณ  BF16 ๋น„์ „ ๊ฐ€์ค‘์น˜๋ฅผ ๊ทธ๋Œ€๋กœ ๋„ฃ์—ˆ์Œ โ†’ ์ด๋ฏธ์ง€ ์ž…๋ ฅ ๋‹ต์ด ํ‹€๋ฆผ(ํ…์ŠคํŠธ ์ž…๋ ฅ์€ ์˜ํ–ฅ ์—†์Œ) โ€” config.json๋งŒ ์ˆ˜์ •, ๋ฌด์‹œ ๋ชฉ๋ก์— vLLM ์ด๋ฆ„ ๊ทœ์น™(re:vpm\..*, re:vit_merger\..*)์„ ์ถ”๊ฐ€ํ•˜๊ณ  ๊ธฐ์กด ๊ทœ์น™ re:model\.visual\..*ยทre:model\.vision_tower\..*ยทre:model\.merger\..*๋ฅผ re:(model\.)?โ€ฆ ํ˜•ํƒœ๋กœ ๋„“ํž˜(๊ฐ€์ค‘์น˜ ๋™์ผ) โ€” vLLM 0.32.0 ํ™•์ธ(2026-11-02): ๋น„์ „ ๊ฐ€์ค‘์น˜๊ฐ€ BF16์œผ๋กœ ๋กœ๋“œ๋จ, ๋ชจ๋ธ์ด ์‹œํ—˜ ๊ทธ๋ฆผ(๋นจ๊ฐ„ ์›ยทํŒŒ๋ž€ ์‚ฌ๊ฐํ˜•ยท๊ธ€์ž "QWEN TEST 2026")์„ ๋ฐ”๋ฅด๊ฒŒ ์„ค๋ช…(vLLM 0.32.0 + transformers 5.21์—์„œ๋Š” ์ด๋ฏธ์ง€ ์ „์ฒ˜๋ฆฌ๊ฐ€ ์‹คํŒจํ•˜๋ฏ€๋กœ โ€” ์š”๊ตฌ์‚ฌํ•ญ ์ฐธ๊ณ  โ€” ์ด ์ ๊ฒ€๋งŒ ์ด๋ฏธ์ง€ ์ฒ˜๋ฆฌ๊ธฐ์— ๋น ์ง„ downsample_mode ์ธ์ž๋ฅผ ๊ธฐ๋ณธ๊ฐ’์œผ๋กœ ์ฑ„์›Œ ์‹คํ–‰) โ€” 2026-11-01 23:41:44 UTC(๋ฆฌ๋น„์ „ a1f77bb) ์ด์ „์— ๋ฐ›์•˜๋‹ค๋ฉด config.json๋งŒ ๋‹ค์‹œ ๋ฐ›์„ ๊ฒƒ

๋ผ์ด์„ ์Šค

  • ๋ผ์ด์„ ์Šค: Apache-2.0 โ€” ์›๋ณธ ๋ชจ๋ธ์˜ ๋ผ์ด์„ ์Šค๋ฅผ ๊ทธ๋Œ€๋กœ ๋”ฐ๋ฆ„
  • ๊ถŒ๋ฆฌ: ์›๋ณธ ๊ฐ€์ค‘์น˜ยทํ† ํฌ๋‚˜์ด์ €๋Š” OpenBMB ์†Œ์œ , ์ด ๋ฆฌํฌ๋Š” ์–‘์žํ™” ํŒŒ์ƒ๋ฌผ
Downloads last month
196
Safetensors
Model size
35B params
Tensor type
BF16
ยท
F8_E4M3
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic

Quantized
(74)
this model

Collection including axteam/MiniCPM-V-4.7-35B-A3B-FP8-dynamic