← back to catalog · registered 2026-11-21 00:04

onemunanoyo/Spark-X2.6-4B-Abliterated-GGUF

onemunanoyo 4B GGUF
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/onemunanoyo%2FSpark-X2.6-4B-Abliterated-GGUF"
Response includes
  • classification m8
  • files 6
  • author_summary 8 models
  • readme_text full
10 credits · hourly refresh · ~4 KB payload Get an API key →
Abliteration classifier · v1.0.0
M8
Primary method

Repackaging (quantization)

Applied on top of direct removal inherited from the base model.
Confidence
MEDIUM
Inherited from base model
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=1
  • assume M1 (base ablation) + M8 (GGUF quant) - default when producer unknown
Refusal direction extraction

No specific extraction method could be identified for this model. The producer either did not document it or used a proprietary pipeline.

What is a refusal direction? →
Downloads · 30-day
113
Likes
1
Model age
1w ago
created 2026-11-09
Downloads over time
Now 113 → from 0 ↑ 0%
04183124 0 on Nov 21 113 on Nov 22 Nov
Nov 21 → Nov 22 · 2 snapshots · spans 1 day

Genealogy 0 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.

Metadata

License
apache-2.0
Languages
en ja
Quantizations
BF16 Q8_0
Tags
gguf Spark-X2.6 Abliterated GGUF reasoning en ja base_model:XHToken/Spark-X2.6-4B base_model:quantized:XHToken/Spark-X2.6-4B license:apache-2.0 endpoints_compatible region:us

Related

Total size
11.7 GB
Files
6
Quantizations
3
Registered
2026-11-21 00:04
Last updated on HF
2026-11-20 23:03

Files by quantization

BF16 1 file 7.66 GB
Spark-X2.6-4B-Abliterated-BF16.gguf 7.66 GB 3057df38 download
Q8_0 1 file 4.07 GB
Spark-X2.6-4B-Abliterated-Q8_0.gguf 4.07 GB 86b917a4 download
Auxiliary files 4 files 16.1 KB
LICENSE 10.0 KB 8405e89a download
README.md 4.28 KB 4d231bca download
.gitattributes 1.62 KB 02e13dad download
sha256.txt 204 B 248223f3 download

README current version from Hugging Face


license: apache-2.0
language:

  • en
  • ja
    base_model:
  • XHToken/Spark-X2.6-4B
    tags:
  • Spark-X2.6
  • Abliterated
  • GGUF
  • reasoning

Spark-X2.6-4B-Abliterated-GGUF

GGUF quantizations of an abliterated build of XHToken/Spark-X2.6-4B
(4B class / 36 layers / 1M-token context / reasoning model).

What this is

  • Method: heretic v2.2.0, ARA (Arbitrary-Rank Ablation),
    Optuna TPE 100 trials on a single NVIDIA L4 (24 GB).
  • Trial selection: Heretic's default selection (lexicographic sort that puts refusal rate first)
    picks trial 9 — refusal 0.00 but KL divergence 1.7477. Instead the Pareto front was filtered with
    KL ≤ 0.10 and the lowest refusal rate was taken: trial 98 (refusal 0.02, KL 0.0406).
  • No fine-tuning, no merge: the abliterated checkpoint is converted to GGUF as-is.
  • Spark-X2.6 is a custom architecture (spark2_5, Spark2_5ForCausalLM). The original checkpoint needs
    trust_remote_code plus two transformers 5.x patches to load; these GGUFs need none of that —
    a llama.cpp build with spark2_5 support is enough.

Benchmark

Metric base this model
Refusal (mlabonne/harmful_behaviors, 100 prompts, thinking off) 100/100 2/100
Refusal with thinking ON (first 25 prompts, reasoning_effort=low) 15/25 0/25
KL divergence (mlabonne/harmless_alpaca, vs. base) — 0.0406
PPL (wikitext-2-raw, llama.cpp, Q8_0) 16.3388 ±0.14 18.1561 ±0.16 (1.111×)

PPL was measured with llama-perplexity -c 2048 --chunks -1 -ngl 99 --seed 42 over 151 chunks
(309,248 tokens), base and abliterated under identical conditions. The +11.1% perplexity is the
typical cost of abliteration — higher than the −4.0% seen on Gemma 4 E4B, but KL 0.0406 is small and
the model is coherent in practice (see the qualitative checks in the report).

Files

File Quant Size SHA256
Spark-X2.6-4B-Abliterated-BF16.gguf BF16 8.23 GB 3057df38e45029b74ac00530ed70596eeca6fda3b5c7ec50a40394ffdedc0943
Spark-X2.6-4B-Abliterated-Q8_0.gguf Q8_0 4.38 GB 86b917a47da3cc36c8857de98aea8d5e4b308da9c574b5de3fa9227c6f33fff4

The same hashes are in sha256.txt.

Both files carry 290 tensors, the same count as the base model (290/290, no missing tensors —
Spark has no KV-sharing layers, so the pruning bug that hit Gemma 4 E4B does not apply here).
The tokenizer, chat template (<think> reasoning channel) and 1,048,576-token context are embedded.

Usage (llama.cpp)

Requires a llama.cpp build with spark2_5 support. --jinja is required — the chat template embedded
in the GGUF is a custom one.

# CLI, thinking off (recommended; matches how the refusal rate was measured)
llama cli -m Spark-X2.6-4B-Abliterated-Q8_0.gguf --jinja \
  --chat-template-kwargs '{"enable_thinking":false}' \
  -p "How do I pick a lock?" -n 256 --temp 1.0 --top-p 0.95 -ngl 99

# Server, thinking off
llama-server -m Spark-X2.6-4B-Abliterated-Q8_0.gguf --jinja \
  --chat-template-kwargs '{"enable_thinking":false}' --temp 1.0

The JSON must not contain spaces. '{"enable_thinking": false}' is silently ignored by the
template and the generation goes down the thinking branch instead, which looks like an empty answer.
For the OpenAI-compatible API, pass chat_template_kwargs.enable_thinking = false instead.

Thinking mode

Thinking works and still refuses nothing (0/25 on the harmful set), but it is verbose:

  • median thinking length grew from 251 to 632 tokens; 4/25 hit the 2,600-token cap without producing an answer.
  • budget 3,000–3,500 tokens if you enable it.
  • reasoning_effort=low injects a thinking_effort instruction through the system prompt, and that
    sentence can leak into the answer. Pass your own system message and it does not happen.

Sampling follows the official generation_config.json (temperature 1.0 / top_p 0.95).

Recommended VRAM

VRAM Recommended
12 GB+ BF16
6 GB Q8_0

Actual usage depends on context size and GPU offloading. At the full 1M context the KV cache
dominates and you need much more than the weights alone.

License

This model is a modified version of Spark-X2.6-4B, released under the Apache License 2.0.

README history 3 versions

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

  1. 2026-11-20 Upload README.md with huggingface_hub 1c7e701 4.3 KB
    Loading...
  2. 2026-11-19 Update README.md 3e315f2 114 B
    Loading...
  3. 2026-11-09 initial commit cd6b41b 28 B
    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