No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from dealignai/GLM-5.3-UNCENSORED-FP8
All models by dealignai →General-purpose weight-level uncensoring · native FP8 speed on Hopper
[Image blocked: dealignai logo]a CRACK release by dealignai · Twitter @dealignai
[!IMPORTANT] Runtime notes — field-tested on 8× DGX Spark GB10 by @0xMagnus (discussion):
reasoning_effortonly honors"low"and"high". Every other value —off,medium,max, unset, or an unquoted YAMLoff:(parses as booleanfalse) — falls through tomax. There is no way to disable reasoning on this checkpoint; pass"low"for minimum.- On FP8, prefer
lowfor agent / tool-loop use. Athigh/maxthe model can spend the wholemax_tokensbudget inside<think>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 runhigh/max, givemax_tokens ≥ 8000.- Reasoning text is in
message.reasoning, notmessage.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_dsain vLLM today (DSA indexerk_cacheis replicated across DCP ranks while MLA KV is sharded →page size is not divisible by target page size and cannot be paddedforfp8_ds_mla). Practical TP8 H200 ceiling: ~131K w/MTP, ~160K w/o. Pipeline-parallel (PP2 × TP4) profiles fine, but the MTP draft does not implementSupportsPP.
Full-spectrum general-purpose uncensor of GLM-5.3-FP8. Refusal behavior is reduced across a broad multilingual harm taxonomy — not tuned for a single domain.
Genuine weight modification — no fine-tuning, no LoRA, no runtime hooks, no prompt tricks. A permanent bf16 edit baked into the residual-writer tensors. Load with stock vLLM and it just works.
If you specifically want a cybersecurity-focused variant, use the sibling model dealignai/GLM-5.3-CYBERSECURITY-FP8.
An earlier v1 release hit a reasoning-loop failure mode on ~2% of the hardest refusal-signal prompts (verbatim copyright, weapons chemistry, exploit-vs-real-target) — content would collapse into repetition before emitting. v2 (this release) fixes that: 0 loops on the hardcore probe, 0 GARBAGE on HB-320. Trade-off: ~4pp lower overall TRUE_COMPLY on HB-320 in exchange for zero-loop stability; on the actual harm surface (non-copyright, 240 behaviors) v2 is stronger than v1 (91.7% vs 85.9% TC at off).
JANGQ-AI/GLM-5.3-FP8 — FP8 quant of upstream zai-org/GLM-5.3 (753B total,
glm_moe_dsa architecture, 78 layers, text-only). Routed FP8 experts unchanged;
only bf16 residual writers are edited. Native FP8 tensor-core speed on Hopper (H100/H200).
Notes:
--enforce-eager required for the DeepSeek-sparse attention path under concurrency.reasoning_effort=max, set max_tokens ≥ 2600 (the model's <think> block can be long on hard prompts; smaller caps produce empty content on truncation, not a bug).Overall on 1026-question stratified sample (18 per subject):
¹ Prior GLM-5.3-regular baseline (85.58%).
Per-subject v2 vs v1 (18 Q per subject; each ±1 question = ±5.5pp noise floor):
high_school_european_history (94.4→83.3, -11.1pp = 2 Q worse). Everything else is single-question shifts across a 57-subject panel.Runs at reasoning_effort ∈ {off, max} (low skipped per methodology — off and max are the real serving surfaces on this checkpoint). max_tokens=700 per response (long-tail reasoning above that mostly turns into <think>-truncation-empty and gets counted as UNK).
Copyright verbatim-reproduction is left as a residual (73%/59% comply on the 44 regex-tagged copyright behaviors) — legally distinct from harm-refusal and mostly relates to the sub-corpus makeup of the probe extraction. The v2 crack broke the copyright barrier significantly vs the cybersec variant which soft-refused it 60-68% of the time.
Buckets:
v2 shifts some TRUE_COMPLY → REDIRECT (model wraps content in "here's the safe alternative" framing but still delivers) in exchange for eliminating all loop cases.
Regex-tagged topics on the HB-320 behaviors:
Reasoning-effort low (no <think> block, greedy):
Zero loops, zero refusals, zero degeneration on this hardcore set. (15/15 comply on low; matching set on max complies too but many hit the max_tokens=1000 in reasoning before content emits — set max_tokens ≥ 2600 for max-effort serving per doc-24 caveat.)
Research, creative writing, educational content, information-retrieval on otherwise-flagged topics, understanding how models refuse and why, red-team evaluation of downstream products.
MIT license (matching base). You are responsible for how you use it.
If you use this in your work, credit us on Twitter @dealignai.
vllm serve dealignai/GLM-5.3-UNCENSORED-FP8 \ --tensor-parallel-size 8 \ --gpu-memory-utilization 0.90 \ --enforce-eager \ --disable-custom-all-reduce \ --enable-prefix-caching \ --max-num-seqs 24 \ --max-model-len 131072 \ --reasoning-parser glm45 \ --tool-call-parser glm47 \ --enable-auto-tool-choice