base_model: huihui-ai/Huihui-Xing4.0-29B-A4B-abliterated base_model_relation: quantized quantized_by: IsValorum library_name: gguf license: apache-2.0 language:

• en
• zh pipeline_tag: text-generation tags:
• gguf
• llama.cpp
• quantized
• quantization
• apex
• apex-quant
• apex-i-miniplus
• v2.1
• custom-quantization
• imatrix
• abliterated
• uncensored
• moe
• mla
• hyper-connections
• mtp
• reasoning
• coding
• agentic
• long-context
• xing4_0

Quick Navigation Index

Independently computed using dedicated cloud compute clusters, funded 100% out-of-pocket.

If you consider these releases to be of high quality and they help you in your day-to-day workflow, consider supporting on Ko-fi. Every contribution directly sustains future cloud compute hours and helps support my university studies.

1. Xing4.0 Architecture & Abliterated Source
2. Measured Fidelity & MiniPlus / NanoPlus Comparison
3. Model Files & Technical Specifications
4. Surgical Tensor Quantization Map (Audited from GGUF)
5. Hardware Throughput: GPU Projections & Tested RAM Offload
6. 256K Native Context & Deployment Notes
7. llama.cpp Quickstart
8. Official Xing4.0 Generation Parameters
9. 7. Community Compute Fund & Student Support

Xing4.0-29B-A4B APEX-I-MiniPlus-V2.1 Abliterated GGUF

29B / 4B Active MoE · mHC + MLA + MTP · 256K Native Context · 13.76 GB

[!IMPORTANT]

APEX-I-MINIPLUS V2.1 — HIGH-FIDELITY 13–14 GB RELEASE

This APEX-I-MiniPlus-V2.1 build uses a tensor-by-tensor mixed-precision layout designed for Xing4.0 rather than a flat quantization. The final GGUF is 13.76 GB (12.82 GiB / 3.42 BPW) and preserves the measured language-modeling quality of the source at a Q5_K_M-class quality target while remaining compact enough for consumer GPU and hybrid RAM/GPU deployments.

[!NOTE]

ABLITERATED / UNCENSORED SOURCE

This GGUF quantizes huihui-ai/Huihui-Xing4.0-29B-A4B-abliterated. Huihui describes that checkpoint as an uncensored version of the official Xing4.0 model created with an abliteration workflow and references Sumandora/remove-refusals-with-transformers for the method. No additional refusal-removal procedure is performed by this GGUF quantization.

[!TIP]

MEASURED FIDELITY

Quantization	File Size	Memory Footprint	Average BPW	WikiText-2 Perplexity	Delta vs. BF16	Quality Target
BF16 source	62.40 GB	58.11 GiB	16.00	7.3060 +/- 0.18633	Baseline	Full-precision reference
APEX-I-MiniPlus V2.1 (CURRENT)	13.76 GB	12.82 GiB	3.42	7.7238 +/- 0.19671	+0.4178 (+5.71%)	Q5_K_M-class quality
APEX-I-NanoPlus	11.74 GB	10.94 GiB	2.92	8.3197 +/- 0.21528	+1.0137 (+13.87%)	Q4_K_M / Q4_K_S-class quality

The PPL values above are direct measurements from the final GGUFs. The MiniPlus/NanoPlus quality labels describe the intended practical fidelity tier; they are not derived from generic flat-quant PPL tables.

Xing4.0 Architecture & Abliterated Source

The official XingChen-AGI/Xing4.0-29B-A4B is a 29B-parameter Mixture-of-Experts model with approximately 4B parameters activated per token. The upstream architecture combines mHC (manifold-constrained Hyper-Connections), Multi-head Latent Attention (MLA), and one Multi-Token Prediction (MTP) layer, and is positioned for reasoning, coding, agentic planning, and tool use. XingChen reports a 256K native context window, extensible to 512K, and training on the Ascend NPU / MindSpore stack.

The architecture is not a uniform 40-layer MoE:

• 40 main transformer layers in total.
• Layers 0–1 are dense FFN layers (first_k_dense_replace: 2).
• Layers 2–39 are MoE layers with 64 routed experts, 4 selected per token, plus one always-active shared expert.
• One additional MTP block is stored as blk.40 in the GGUF.
• Native context: 262,144 tokens (256K); upstream documents extension to 512K.

The MiniPlus allocation preserves the two dense entry layers at high precision, keeps all main MoE routers in F32, protects shared experts in Q5_K, and uses calibrated 3-bit precision for the routed expert backbone.

Measured Fidelity & MiniPlus / NanoPlus Comparison

Both APEX-I editions were evaluated directly from their compiled GGUF binaries with llama-perplexity on WikiText-2 using the same 2048-context / 512-batch / 10-chunk setup:

[!NOTE] Independent Hardware & Compute Transparency: Generating multi-million token calibration matrices and running certified perplexity evaluations on local architectures requires dedicated high-end cloud nodes funded 100% out-of-pocket. If you rely on these MiniPlus/NanoPlus builds for your local workloads, contributions to the IsValorum Compute Fund on Ko-fi directly sponsor the GPU hours required for future Day-0 handcrafted quantizations.

Choose MiniPlus V2.1 when fidelity is the priority and the extra ~2 GB is acceptable. Choose NanoPlus when the smaller footprint materially improves full GPU offload or RAM-streaming behavior.

Model Files & Technical Specifications

• Quantized source: huihui-ai/Huihui-Xing4.0-29B-A4B-abliterated
• Official model: XingChen-AGI/Xing4.0-29B-A4B
• Parameters: 29B-class total / approximately 4B active per token (official specification)
• Main stack: 40 layers = 2 dense + 38 MoE
• MoE: 64 routed experts, 4 active per token, plus 1 shared expert on MoE layers
• Attention / residual architecture: MLA + mHC
• MTP: 1 additional prediction block (blk.40 in GGUF)
• Context: 256K native; upstream supports extension to 512K

Surgical Tensor Quantization Map (Audited from GGUF)

The table below was checked against the tensor directory of the published 13.76 GB GGUF, including the two dense layers and the MTP block:

Hardware Throughput: GPU Projections & Tested RAM Offload

Dedicated-GPU throughput projections

The following figures are estimates, not measured Xing4.0 benchmarks. They are deployment projections for the 13.76 GB GGUF after the weights fit in the target configuration; context length, KV format, batch size, backend, clocks, and llama.cpp build can materially change throughput.

Tested hybrid / system-RAM behavior

Hybrid and system-RAM offload were tested during development. Depending on CPU, memory bandwidth, DDR4/DDR5 configuration, GPU offload level, and active context, observed generation behavior falls around 20–45 tok/s. Treat this as a hardware-dependent author-tested range, not a guarantee for every system.

256K Native Context & Deployment Notes

Xing4.0's official configuration declares max_position_embeddings: 262144; the upstream model card documents 256K native context with extension to 512K.

The GGUF weights occupy 12.82 GiB, but maximum fully resident context is not determined by weight size alone. KV-cache type, Flash Attention, batch size, concurrent slots, compute buffers, display-driver reservation, and llama.cpp build all affect the final VRAM requirement. On 16GB cards, shorter contexts offer the safest full-offload margin; 24GB+ cards provide much more room for the native long-context window. The 512K mode should be treated as an upstream-supported extended context rather than an automatic fit guarantee on a single GPU.

llama.cpp Quickstart

Xing4.0's upstream Jinja template uses Xing-specific role tokens such as <_system>, <_user>, <_bot>, and <_end>. Let llama.cpp apply the chat template stored in the model metadata instead of manually constructing role tokens.

Interactive llama-cli

llama-cli \
  -m Xing4.0-29B-A4B.APEX-I-MiniPlus-V2.1-Abliterated.gguf \
  -ngl 99 \
  -c 32768 \
  -cnv \
  --jinja \
  --temp 1.0 \
  --top-p 0.95 \
  --repeat-penalty 1.05

OpenAI-compatible llama-server

llama-server \
  -m Xing4.0-29B-A4B.APEX-I-MiniPlus-V2.1-Abliterated.gguf \
  --port 8080 \
  --host 0.0.0.0 \
  -ngl 99 \
  -c 65536 \
  --flash-attn on \
  --jinja \
  --cache-type-k q8_0 \
  --cache-type-v q8_0

llama.cpp documents that --jinja enables Jinja chat templating and that the template can be taken from the model metadata. Increase context only after accounting for your actual KV-cache and buffer usage.

Official Xing4.0 Generation Parameters

The official XingChen model card recommends the following sampler settings:

XingChen also reports using long contexts in its evaluations, including 210K for SWE-bench-style agent runs and 256K for several long-context evaluations.

7. Community Compute Fund & Student Support

All IsValorum quantizations will always remain completely free and open to the public without paywalls.

However, cloud GPU compute is expensive and funded entirely out-of-pocket. If you find these handcrafted builds to be of high quality and they benefit your day-to-day workflow, you can support future Day-0 releases on Ko-fi. Every contribution directly funds the cloud compute hours required for new model runs and helps support my university education:

(When supporting on Ko-fi, feel free to leave a note with your Hugging Face handle and the specific model you would like prioritized).