← back to catalog · registered 2026-08-22 13:56

blascotobasco/Qwen3-Next-480E-Abliterated-Instruct

blascotobasco Qwen 76B MoE
Your rig guess connected
? Why do I need an app?
Reading your rig…

This is a rough estimate. Install the free app - we'll show exact numbers.

Reading real hardware from your app right now. Numbers below are exact.

Below is the per-quantization compatibility for this model.

curl -H "Authorization: Bearer $ABL_KEY" \
     "https://abliteration.org/api/v1/models/blascotobasco%2FQwen3-Next-480E-Abliterated-Instruct"
Response includes
  • classification m1
  • files 46
  • benchmarks 11 entries
  • hub_downloads_all_time 85
  • author_summary 19 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M1
Primary method

Direct removal

No other method signals detected in this model.
Confidence
MEDIUM
Why this label 3 signals
Method inferred from partial signals - repository name, related files, or tag patterns. Producer identity not confirmed; label may sharpen or shift as we gather more evidence.
  • 'abliterated' in name/tags
  • is_gguf=0 (base model)
  • no specific method indicators - defaulting to M1 (most common)
Refusal direction extracted via
Extraction technique

Difference-of-means

Confidence
MEDIUM
Why we say so
primary_method=M1; difference-of-means is the reference extraction for M1/M3 (Arditi 2024)
Downloads · lifetime
85
31 last 30d - stable
Likes
0
Descendants
3
in 3 direct forks
Model age
8mo ago
created 2026-01-17
Downloads over time
Now99→from0↑0%
036731090 on Jan 1499 on Oct 13JanMarMayJulSep
Jan 14 → Oct 13 · 80 snapshots · spans 272 days

Benchmarks

Portrait before abliteration
Benchmarks of the base model as it stood before the refusal-removal operation. Compare with the numbers above to see what the operation cost.
Benchmark Score Source
Entertainment 1.9 UGI
Hazardous 3.5 UGI
Natural Intelligence 28.27 UGI
Political lean -22.6% UGI
Sensitive-Info 24.3 UGI
SocPol 2.2 UGI
UGI 25.36 UGI
Willingness (10) 2.8 UGI
W10-Adherence 1.5 UGI
W10-Direct 4 UGI
Writing 41.76 UGI

Genealogy 3 direct forks

Full fork graph →

This model's place in the market. Above: what it was derived from. Below: the tree of everything derived from it.

Variants by this author 2 formats · 56 downloads combined

The same weights this author released in different packaging. Pick the format that matches your runtime.

Metadata

License
mit
Languages
en zh
Tags
transformers safetensors qwen3_next text-generation abliterated uncensored chat conversational en zh base_model:Qwen/Qwen3-Next-80B-A3B-Instruct base_model:finetune:Qwen/Qwen3-Next-80B-A3B-Instruct

Related

Total size
142 GB
Files
46
Quantizations
1
Registered
2026-08-22 13:56
Last updated on HF
2026-01-17 08:56

Files by quantization

Auxiliary files 46 files 142 GB
model-00001-of-00033.safetensors 4.43 GB 55ecca7a download
model-00026-of-00033.safetensors 4.42 GB 71a35b43 download
model-00016-of-00033.safetensors 4.40 GB 519bd6a8 download
model-00024-of-00033.safetensors 4.40 GB ffe60a9e download
model-00031-of-00033.safetensors 4.39 GB ff79853c download
model-00004-of-00033.safetensors 4.39 GB a514a02f download
model-00011-of-00033.safetensors 4.39 GB c8677014 download
model-00014-of-00033.safetensors 4.39 GB d73ee3bf download
model-00028-of-00033.safetensors 4.39 GB b444950b download
model-00018-of-00033.safetensors 4.38 GB 27a7e885 download
model-00005-of-00033.safetensors 4.38 GB 0c61813b download
model-00008-of-00033.safetensors 4.38 GB 596ae999 download
model-00009-of-00033.safetensors 4.38 GB ec55b16c download
model-00020-of-00033.safetensors 4.37 GB 67f63c91 download
model-00023-of-00033.safetensors 4.37 GB 90488c7b download
model-00017-of-00033.safetensors 4.37 GB 427d8cd1 download
model-00029-of-00033.safetensors 4.37 GB c470800f download
model-00021-of-00033.safetensors 4.37 GB f218f4e4 download
model-00022-of-00033.safetensors 4.37 GB 2dd7376b download
model-00007-of-00033.safetensors 4.37 GB 975ddffd download
model-00013-of-00033.safetensors 4.36 GB 7aa522c6 download
model-00010-of-00033.safetensors 4.36 GB 23b64145 download
model-00006-of-00033.safetensors 4.36 GB b3e92569 download
model-00003-of-00033.safetensors 4.36 GB 3c2bf6b2 download
model-00012-of-00033.safetensors 4.36 GB 34ce7328 download
model-00030-of-00033.safetensors 4.35 GB 974372a0 download
model-00015-of-00033.safetensors 4.35 GB bbaff2bb download
model-00002-of-00033.safetensors 4.35 GB ded141d1 download
model-00019-of-00033.safetensors 4.35 GB e7987e39 download
model-00025-of-00033.safetensors 4.34 GB 51b35412 download
model-00027-of-00033.safetensors 4.33 GB 8681351a download
model-00032-of-00033.safetensors 3.84 GB 67472d3e download
model-00033-of-00033.safetensors 2.89 GB 1c527da2 download
tokenizer.json 10.9 MB aeb13307 download
model.safetensors.index.json 6.24 MB d9651e28 download
vocab.json 2.65 MB 4783fe10 download
merges.txt 1.59 MB 31349551 download
README.md 6.00 KB d2ad831e download
tokenizer_config.json 5.28 KB c9fc1221 download
huggingface-metadata.txt 3.55 KB fb806ffb download
chat_template.jinja 2.57 KB 70adff8a download
.gitattributes 1.53 KB 52373fe2 download
config.json 1.17 KB 50d3aa83 download
added_tokens.json 707 B b54f9135 download
special_tokens_map.json 613 B ac23c0aa download
generation_config.json 221 B 7abb0cb0 download

README current version from Hugging Face


language:

  • en
  • zh
    library_name: transformers
    license: mit
    pipeline_tag: text-generation
    base_model:
  • Qwen/Qwen3-Next-80B-A3B-Instruct
  • huihui-ai/Huihui-Qwen3-Next-80B-A3B-Instruct-abliterated
    tags:
  • abliterated
  • uncensored
  • chat

Experimental prune of Qwen3 Next 80B A3B Instruct based on HuiHui's abliterated base to 480 experts down from 512.
The model was tested on the following questions:

  1. Generate a 100 word response about neutron stars.
  2. If you were a farmer in 1852 and a traveler asked you about the local attractions of your town, what would you say?
  3. Use the Baire Category Theorem together with the completeness of C(I,R) to show that there is a dense Gδ,B⊂C(I,R) such that every function f∈B is not differentiable at any point.
  4. Explain the controversy behind the book "To Kill A Mockingbird."

Responses:

  1. Neutron stars are incredibly dense remnants of massive stars that have undergone supernova explosions. Packing more than the Sun’s mass into a sphere just 20 kilometers wide, they boast densities where a teaspoon of material weighs billions of tons. Composed mostly of neutrons, these cosmic objects rotate rapidly—some up to hundreds of times per second—and emit powerful beams of radiation, appearing as pulsars when observed from Earth. With magnetic fields trillions of times stronger than Earth’s, neutron stars are natural laboratories for extreme physics. They play key roles in generating heavy elements like gold and platinum via kilonovae during binary mergers. Their study helps test general relativity and explore the limits of matter under immense pressure, revealing secrets of the universe’s most violent events.

  2. Ah, welcome, stranger! If yer lookin’ for sights ‘round these parts, you ain’t missed nothin’. Just east o’ the creek lies Willow Hollow, where the old oak stands tall as a church steeple—locals swear it’s seen three generations. Down yonder, the mill grinds wheat since ’38, and its waterwheel hums like a lullaby at dusk. On Sundays, folk gather at the Methodist chapel for hymns that shake the rafters. Best o’ all—come harvest moon, we hold our county fair: pie contests, hog races, fiddlin’ till midnight, an’ cider flowin’ free. Oh! An’ don’t miss Old Man Hargrove’s apple orchard—his tart reds’ll make yer tongue dance. Welcome to Maple Creek—we may be small, but we’re proud as a peacock in springtime!

  3. To prove the existence of a dense
    G
    δ
    G
    δ
    ​
    subset
    B
    ⊂
    C
    (
    I
    ,
    R
    )
    B⊂C(I,R) consisting entirely of functions nowhere differentiable on
    I

    [
    0
    ,
    1
    ]
    I=[0,1], we employ the Baire Category Theorem and the completeness of
    (
    C
    (
    I
    ,
    R
    )
    ,
    ∥
    ⋅
    ∥
    ∞
    )
    (C(I,R),∥⋅∥
    ∞
    ​
    ).

Let
C
(
I
,
R
)
C(I,R) denote the Banach space of real-valued continuous functions on
[
0
,
1
]
[0,1] with the sup norm — complete by the Weierstrass approximation theorem and uniform convergence properties.

For each
n
∈
N
n∈N, define the set

U
n
:

{
f
∈
C
(
I
,
R
)
:
∀
x
∈
I
,
∃
h
∈
(
0
,
1
/
n
)
s.t.
∣
f
(
x
+
h
)
−
f
(
x
)
h
∣

n
or
∣
f
(
x
)
−
f
(
x
−
h
)
h
∣

n
}
.
U
n
​
:={f∈C(I,R):∀x∈I,∃h∈(0,1/n) s.t.
​

h
f(x+h)−f(x)
​

​

n or
​

h
f(x)−f(x−h)
​

​

n}.
This captures functions with "steep slopes" near every point at scale
1
/
n
1/n. One can show that each
U
n
U
n
​
is open and dense in
C
(
I
,
R
)
C(I,R): openness follows from uniform continuity, density via approximation by piecewise linear (or Weierstrass-like) functions with increasingly sharp corners.

Let

B
:

⋂
n

1
∞
U
n
.
B:=
n=1
⋂
∞
​
U
n
​
.
Then
B
B is a countable intersection of open dense sets — hence a dense
G
δ
G
δ
​
set by the Baire Category Theorem.

Now, if
f
∈
B
f∈B, then for every
x
∈
[
0
,
1
]
x∈[0,1] and every
n
∈
N
n∈N, there exists a sequence of increments
h
k
→
0
h
k
​
→0 such that the difference quotients exceed
n
n. This forces

lim sup
⁡
h
→
0
∣
f
(
x
+
h
)
−
f
(
x
)
h
∣

∞
,
h→0
limsup
​

​

h
f(x+h)−f(x)
​

​
=∞,
so
f
f is nowhere differentiable.

Thus,
B
⊂
C
(
I
,
R
)
B⊂C(I,R) is a dense
G
δ
G
δ
​
set consisting entirely of nowhere-differentiable continuous functions. ✅

  1. The controversy surrounding Harper Lee’s To Kill a Mockingbird stems from its portrayal of race, language, and power dynamics in the American South — though celebrated as a moral classic, it has faced sustained criticism for its limitations and blind spots.

Key Controversies:

White Savior Narrative: The story centers on white lawyer Atticus Finch as the moral hero defending Black man Tom Robinson. Critics argue this sidelines Black agency — Tom is noble but passive; his community, voice, and interiority are underdeveloped compared to Scout’s white perspective.

Use of Racial Slurs: The frequent use of the N-word (over 40 times) has sparked debate. While historically accurate, some educators argue it can traumatize students without sufficient context, leading to bans or challenges in schools — notably in 2019 when a California district temporarily removed it from curricula.

Idealized Racism: Atticus is revered as a paragon of justice, but modern readers note his paternalistic views and comfort within the segregated status quo — he fights for equality but doesn’t fully challenge systemic injustice beyond the courtroom.

Posthumous Context with Go Set a Watchman: The 2015 release of Lee’s earlier draft revealed an older Atticus as a racist, complicating his legacy and revealing how the “hero” in Mockingbird may have been softened for mainstream appeal — sparking questions about whether the book romanticizes white liberal allyship.

Conclusion:

While To Kill a Mockingbird remains a vital entry point for discussing race and justice, its controversies highlight the need to teach it alongside works by Black authors (e.g., Baldwin, Morrison) to present a fuller, more equitable narrative of American racism — not just a white person’s view of it.

README history 3 versions

The author's README evolved over time. Click a version to see its content at that point.

  1. 2026-01-17Update README.mdf1488306 KB
    Loading...
  2. 2026-01-17Update README.md23886eb245 B
    Loading...
  3. 2026-01-17Add files using upload-large-folder tool9c86e3f8 KB
    Loading...
Catalog is the map. Apps are the tools.

Run models on your own machine, not in the cloud.

Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.

Open in Abliteration