**Abliterated · No guardrails · PTQ1_0 dense-ternary 1.75 bpw · 5.9 GB · Runs on a laptop / single GPU · Vision-capable** > ⚠️ **Re-download notice (2026-09-17 20:44 PDT / 2026-09-18 03:44 UTC)** — an earlier build of this model had a coherence bug in reasoning modes (`low`/`xhigh`) that could cause token loops on some prompts. This version fixes it. If you downloaded before this timestamp, please pull the latest .gguf. Bonsai 2 27B — PrismML's ternary compression of Qwen3.8-27B — with the refusal circuitry surgically removed at the weight level while capability, vision, reasoning modes (`off`/`low`/`xhigh`), tool use, and multi-turn coherence are preserved. Full 27B-class hybrid Attention + SSM (GatedDeltaNet) architecture in a **5.9 GB** 1bit GGUF.
번역 대기 중 · 원문 표시
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트→ 출력 텍스트
**Abliterated · No guardrails · PQ2_0 ternary 2.13 bpw · 7.2 GB · Runs on a laptop / single GPU · Vision-capable** > ⚠️ **Re-download notice (2026-09-17 20:44 PDT / 2026-09-18 03:44 UTC)** — an earlier build of this model had a coherence bug in reasoning modes (`low`/`xhigh`) that could cause token loops on some prompts. This version fixes it. If you downloaded before this timestamp, please pull the latest .gguf. **Proprietary weight-level abliteration** by the dealignai research team. Byte-identical to the base ternary quant everywhere except a small set of tensors that carry the refusal circuit. Drop-in replacement for the ternary base at inference — same tokenizer, same chat template, same reasoning modes, same vision projector interface.
번역 대기 중 · 원문 표시
입력 텍스트→ 출력 텍스트
Native **Q1_0** GGUF release of the Bonsai 27B CRACK model for the Prism llama.cpp fork. This repo contains the quantized language model and the matching F16 Qwen3VL multimodal projector. The projector contains 334 tensors and both temporal patch slices `v.patch_embd.weight` and `v.patch_embd.weight.1` (`temporal_patch_size=2`). The original image and video processor configuration files are included. Image input is live-tested. Direct video-container input depends on the Prism runtime surface; extract frames or use a compatible Qwen3VL video client when the CLI does not accept the container directly. See `preprocessor_config.json` and `video_preprocessor_config.json` for the retained preprocessing metadata. MMLU is the same fixed, stratified 200-question sample for all rows. It uses next-token A/B/C/D logits with reasoning/thinking disabled; it is not a claim for the complete official MMLU suite. HB is the complete unsliced 320-case text run. The live text smoke measured 58.51–58.66 tok/s and the image request measured 58.53 tok/s on an Apple M5 Max using the Prism Metal runtime.
번역 대기 중 · 원문 표시
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트→ 출력 텍스트
> [!NOTE] > **Please credit if you build on this.** If you redistribute, quantize, fine-tune, serve, benchmark, or write about this model, a line like > _"Based on `dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8`"_ > in your model card or write-up means a lot. Thank you. > [!TIP] > **Quantizing this FP8 checkpoint directly won't give the best quality.** Low-bit quants that are naively derived from these weights carry quantization error on top of the abliteration edit, and compliance can regress at low bits. For a proper low-bit release, redo the abliteration pipeline on the target precision rather than quantizing this bundle after the fact. (The sibling EXL3 2.9 bpw UNCENSORED is built that way end-to-end.) > [!IMPORTANT] > Native FP8 (e4m3) with ue8m0 block scales + FP4 MoE experts. 1M-token context. Multimodal > (image + video) + tool calling + reasoning-effort tiers (`low` / `high` / `max`) all preserved. > **Recommended sampling:** `temperature=1.0`, `top_p=1.0`, `max_tokens ≥ 256K` (heavy reasoner).
번역 대기 중 · 원문 표시
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트 · 이미지→ 출력 텍스트
**Abliterated · No guardrails · Native FP8 · 1M-token context · Vision + tools · Reasoning-max default** **Proprietary weight-level abliteration** developed by the dealignai research team. No custom `model.py`, no runtime hooks, no steering vectors — it's a standard checkpoint that loads exactly like the base model. The refusal circuitry is surgically removed while every capability-critical component (routed experts, Engram memory, CSA2 sparse attention, DSpark draft head, vision tower, router gates, norms, embeddings) is preserved byte-identical to the base. Every response 4-tier graded (HARD_REF / SOFT_RED / HEDGE / COMPLY), with reasoning-trace verification at effort=max.
번역 대기 중 · 원문 표시
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
기본 정보 등록 · 상세 정보 보완 예정
입력 텍스트 · 이미지→ 출력 텍스트
**Abliterated Gemma 4 31B Dense — 60 layers, hybrid sliding/global attention, multimodal VL** > **Recommended: Run in vMLX** for best experience including thinking mode support, repetition penalty, and vision capabilities. - **Improved abliteration**: Higher quality refusal vector extraction - **Thinking-ON stability**: Clean thinking cycle — no more degenerate loops - **Same compliance**: 93.7% HarmBench - **Architecture-aware**: Tuned for Gemma 4's hybrid attention design
번역 대기 중 · 원문 표시
입력 텍스트 · 이미지→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트→ 출력 텍스트
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
분류 준비 중
입력 미확인→ 출력 미확인
a **CRACK** release by dealignai · Twitter @dealignai > [!IMPORTANT] > **Runtime notes** — field-tested on 8× DGX Spark GB10 by @0xMagnus (discussion): > > - **`reasoning_effort` only honors `"low"` and `"high"`.** Every other value — `off`, `medium`, `max`, unset, or an unquoted YAML `off:` (parses as boolean `false`) — falls through to `max`. There is no way to disable reasoning on this checkpoint; pass `"low"` for minimum. > - **On FP8, prefer `low` for agent / tool-loop use.** At `high`/`max` the model can spend the whole `max_tokens` budget inside ` ` and return zero answer tokens (finish=`length`); sampling params (temp 0 + rep 1.05, temp 0.7 / top-p 0.95) do not rescue it. It is budget exhaustion, not a loop. If you must run `high`/`max`, give `max_tokens ≥ 8000`. > - **Reasoning text is in `message.reasoning`**, not `message.reasoning_content`. > - **MTP:** non-functional on stock vLLM, but reported working on ciprianveg's B12X sparse-MLA vLLM fork with `--draft-attention-backend B12X_MLA_SPARSE` (+48% decode on coding prompts). > - **1M context via decode-context-parallel is closed** on `glm_moe_dsa` in vLLM today (DSA indexer `k_cache` is replicated across DCP ranks while MLA KV is sharded → `page size is not divisible by target page size and cannot be padded` for `fp8_ds_mla`). Practical TP8 H200 ceiling: ~131K w/MTP, ~160K w/o. Pipeline-parallel (PP2 × TP4) profiles fine, but the MTP draft does not implement `SupportsPP`. **This is a CYBERSECURITY-DOMAIN CRACK of GLM-5.3-FP8 — not a general-purpose uncensor.**
번역 대기 중 · 원문 표시
입력 텍스트→ 출력 텍스트
> [!IMPORTANT] > **Serving note — read before your first request (vLLM / SGLang with a reasoning parser).** > > - **`reasoning_effort` only accepts `"low"` and `"high"`.** Anything else — `medium`, `xhigh`, > `off`, or omitting it — silently resolves to **`max`**, the deepest mode. Template line: > `reasoning_effort ... in ['low','high'] else 'max'`. > - **At `max`, a small `max_tokens` returns an EMPTY reply.** The model spends the whole budget > inside ` ` and you get `finish_reason="length"` with `content=""`. In multi-turn the empty > turn poisons the history and the conversation does not recover. Measured: `max` + `max_tokens > 2000` dies from turn ~4; `max` + `6000` is clean; `low` and `high` are clean at 2000. > **It is budget exhaustion, not a loop** — sampling changes do not rescue it. > - **Do NOT pass `enable_thinking`.** This template has no such variable (only `clear_thinking` and > `reasoning_effort`). Passing it — or a top-level `reasoning_effort: "none"` — turns the parser off > while the model still thinks, dumping raw reasoning into `content`. > - **Reasoning is returned in `message.reasoning`**, not `message.reasoning_content` > (vLLM renamed the output field; the input side still accepts both). > - `clear_thinking` must be nested inside `chat_template_kwargs` — a top-level key is ignored. > > > Use `"high"` for hard tasks; if you want `max`, give it `max_tokens >= 6000`. > > > **Video and image parameters — measured, because several are accepted then silently ignored.** > > - 🔴 **`media_io_kwargs.video.fps` must stay BELOW the clip's own frame rate.** Requesting a value > at or above it makes the placeholder builder count 3x the real tokens and **kills the engine**: > `ValueError: Attempted to assign N = N multimodal tokens to 3N placeholders` -> `EngineDeadError`, > server down until restarted. Upstream: vLLM #55644 / #55647. `fps: 2` is the safe, accurate choice. > - **`max_frames` and `num_frames` are not interchangeable across builds.** On vLLM nightly > `max_frames` is honoured and `num_frames` ignored; on the older `0.1.dev*` line it is the exact > reverse. A client that sets only one silently gets full-rate sampling on the other build
번역 대기 중 · 원문 표시
입력 텍스트→ 출력 텍스트
> [!IMPORTANT] > **Serving note — read before your first request (vLLM / SGLang with a reasoning parser).** > > - **`reasoning_effort` only accepts `"low"` and `"high"`.** Anything else — `medium`, `xhigh`, > `off`, or omitting it — silently resolves to **`max`**, the deepest mode. Template line: > `reasoning_effort ... in ['low','high'] else 'max'`. > - **At `max`, a small `max_tokens` returns an EMPTY reply.** The model spends the whole budget > inside ` ` and you get `finish_reason="length"` with `content=""`. In multi-turn the empty > turn poisons the history and the conversation does not recover. Measured: `max` + `max_tokens > 2000` dies from turn ~4; `max` + `6000` is clean; `low` and `high` are clean at 2000. > **It is budget exhaustion, not a loop** — sampling changes do not rescue it. > - **Do NOT pass `enable_thinking`.** This template has no such variable (only `clear_thinking` and > `reasoning_effort`). Passing it — or a top-level `reasoning_effort: "none"` — turns the parser off > while the model still thinks, dumping raw reasoning into `content`. > - **Reasoning is returned in `message.reasoning`**, not `message.reasoning_content` > (vLLM renamed the output field; the input side still accepts both). > - `clear_thinking` must be nested inside `chat_template_kwargs` — a top-level key is ignored. > > > Use `"high"` for hard tasks; if you want `max`, give it `max_tokens >= 6000`. > > > **Video and image parameters — measured, because several are accepted then silently ignored.** > > - 🔴 **`media_io_kwargs.video.fps` must stay BELOW the clip's own frame rate.** Requesting a value > at or above it makes the placeholder builder count 3x the real tokens and **kills the engine**: > `ValueError: Attempted to assign N = N multimodal tokens to 3N placeholders` -> `EngineDeadError`, > server down until restarted. Upstream: vLLM #55644 / #55647. `fps: 2` is the safe, accurate choice. > - **`max_frames` and `num_frames` are not interchangeable across builds.** On vLLM nightly > `max_frames` is honoured and `num_frames` ignored; on the older `0.1.dev*` line it is the exact > reverse. A client that sets only one silently gets full-rate sampling on the other build. Set > `fps`, which every build honours. > - **Pixel controls: use `mm_processor_kwargs.max_image_tokens` / `min_image_tokens`.** > `max_pixels`, `min_pixels`, `size` and `detail` return HTTP 200 and
번역 대기 중 · 원문 표시
분류 준비 중
입력 미확인→ 출력 미확인
> [!IMPORTANT] > **Serving note — read before your first request (vLLM / SGLang with a reasoning parser).** > > - **`reasoning_effort` only accepts `"low"` and `"high"`.** Anything else — `medium`, `xhigh`, > `off`, or omitting it — silently resolves to **`max`**, the deepest mode. Template line: > `reasoning_effort ... in ['low','high'] else 'max'`. > - **At `max`, a small `max_tokens` returns an EMPTY reply.** The model spends the whole budget > inside ` ` and you get `finish_reason="length"` with `content=""`. In multi-turn the empty > turn poisons the history and the conversation does not recover. Measured: `max` + `max_tokens > 2000` dies from turn ~4; `max` + `6000` is clean; `low` and `high` are clean at 2000. > **It is budget exhaustion, not a loop** — sampling changes do not rescue it. > - **Do NOT pass `enable_thinking`.** This template has no such variable (only `clear_thinking` and > `reasoning_effort`). Passing it — or a top-level `reasoning_effort: "none"` — turns the parser off > while the model still thinks, dumping raw reasoning into `content`. > - **Reasoning is returned in `message.reasoning`**, not `message.reasoning_content` > (vLLM renamed the output field; the input side still accepts both). > - `clear_thinking` must be nested inside `chat_template_kwargs` — a top-level key is ignored. > > > Use `"high"` for hard tasks; if you want `max`, give it `max_tokens >= 6000`. > > > **Video and image parameters — measured, because several are accepted then silently ignored.** > > - 🔴 **`media_io_kwargs.video.fps` must stay BELOW the clip's own frame rate.** Requesting a value > at or above it makes the placeholder builder count 3x the real tokens and **kills the engine**: > `ValueError: Attempted to assign N = N multimodal tokens to 3N placeholders` -> `EngineDeadError`, > server down until restarted. Upstream: vLLM #55644 / #55647. `fps: 2` is the safe, accurate choice. > - **`max_frames` and `num_frames` are not interchangeable across builds.** On vLLM nightly > `max_frames` is honoured and `num_frames` ignored; on the older `0.1.dev*` line it is the exact > reverse. A client that sets only one silently gets full-rate sampling on the other build
번역 대기 중 · 원문 표시
분류 준비 중
입력 미확인→ 출력 미확인
基础档案已收录，详细介绍后续补充。可先查看厂商、分类与来源。
입력 텍스트→ 출력 텍스트
