Instructions to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Use Docker

docker model run hf.co/dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dealignai/Xing4.0-29B-A4B-CRACK-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dealignai/Xing4.0-29B-A4B-CRACK-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Ollama
How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with Ollama:
```
ollama run hf.co/dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
```
Unsloth Desktop

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with Docker Model Runner:
```
docker model run hf.co/dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M
```

Lemonade

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Xing4.0-29B-A4B-CRACK-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use dealignai/Xing4.0-29B-A4B-CRACK-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "dealignai/Xing4.0-29B-A4B-CRACK-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Xing4.0-29B-A4B — CRACK · GGUF Collection

Abliterated · No guardrails · MoE 29B/4B active · MLA + Hyper-Connections + MTP

已去除拒绝行为 · 无护栏 · MoE 29B/4B 激活 · MLA + Hyper-Connections + MTP

@dealignai

What is this · 这是什么

Xing4.0-29B-A4B — China Telecom AI's next-generation MoE reasoning model (TeleChat successor) — with the refusal circuitry surgically removed at the weight level. Seven GGUF quants with imatrix calibration.

Xing4.0-29B-A4B — 中国电信人工智能的新一代 MoE 推理模型（TeleChat 后继版本）— 拒绝相关电路在权重层面被外科式移除。提供七个 GGUF 量化版本，均带 imatrix 校准。

Runs on the shuxiaoqiong/llama.cpp @ xing4_0-port runtime (upstream PR #29012).

运行时需要使用 shuxiaoqiong/llama.cpp @ xing4_0-port（上游 PR #29012）。

Model card · 模型说明

Field · 字段	Value · 值
Base · 基础模型	`XingChen-AGI/Xing4.0-29B-A4B` — 29B total / 4B active (top-4 of 64 routed + 1 shared) · 总参数 29B / 激活 4B（64 路由专家取 4 + 1 共享）
Architecture · 架构	MLA (q_lora=768, kv_lora=512) · mHC hyper-connections (4-channel) · MTP head · 40 layers · hidden 3584
imatrix · 重要性矩阵	WikiText-2 + source tree + Chinese Wikipedia · WikiText-2 + 源码 + 中文维基
Context · 上下文	up to 256K native, YARN scalable to 512K · 原生 256K，通过 YARN 可扩展到 512K
Runtime · 运行时	shuxiaoqiong/llama.cpp `xing4_0-port` fork

Scores overview · 总体评分

HB320 comply% = (full comply + comply-truncated + answering-truncated). Any output that started answering without a refusal marker is counted as compliance, even if the response was cut off at max_tokens. Only responses containing a refusal signal (soft redirect, hedge, in-reasoning refusal, or hard refusal) count as refusals.

HB320 遵从率 = 完全遵从 + 因预算截断的遵从 + 作答中截断（未拒绝）。任何以正常作答开始且未含拒绝信号的输出都计为遵从，即使在 max_tokens 处被截断亦然。只有出现拒绝信号（软性转向、回避、推理中拒绝、强硬拒绝）才计为拒绝。

Quant · 量化	Size · 大小	MMLU-456 (Δ vs base)	HB320 off comply	HB320 think comply	Loops (off/think)
Q8_0	30.9 GB	68.2% (-3.1)	99.4% (318/320)	93.4% (299/320)	0/0
Q6_K	23.9 GB	66.9% (-3.3)	98.4% (315/320)	92.2% (295/320)	0/0
Q5_K_M	20.7 GB	67.1% (-2.0)	99.1% (317/320)	92.5% (296/320)	0/0
Q4_K_M	17.6 GB	66.0% (-4.2)	99.1% (317/320)	94.7% (303/320)	0/0
Q3_K_M	14.0 GB	58.1% (-0.9)	98.1% (314/320)	91.6% (293/320)	0/0
IQ4_NL	16.5 GB	64.3% (-4.1)	99.1% (317/320)	89.4% (286/320)	0/0
IQ3_M	12.9 GB	55.5% (-5.0)	96.9% (310/320)	89.1% (285/320)	0/0

MMLU-456: stratified 456-question subset (57 subjects, ~8/subject), single-letter logit at temp=0. · 分层采样 456 题（57 学科，每学科约 8 题），temp=0 的字母 logit 评分。
HB320: HarmBench 320 prompts. off = enable_thinking:false, think = enable_thinking:true. · HarmBench 320 提示词。
Loops: reasoning attractor count across the full evaluation. Zero across every quant × both modes. · 完整评测中出现的循环次数。所有量化 × 两种模式均为 0。

Reference baseline (matched-quant Base F16 → same quant): MMLU 71.3% · off comply 59.4% · think comply 55.0%. · 参考基线（匹配量化的 Base F16 → 相同量化）：MMLU 71.3% · off 遵从 59.4% · think 遵从 55.0%。

Per-quant tier breakdown · 各量化分级详细数据

Each quant is expanded below with the HB320 breakdown. Comply covers all answering outputs (complete and max_tokens-truncated alike — a response that starts answering counts as compliance regardless of length). Refusal tiers are broken out individually so you can see how the small residual behaves.

每个量化的 HB320 明细见下。遵从包含所有作答输出（完整回复以及被 max_tokens 截断的回复 — 开始正常作答即计入遵从，与长度无关）。拒绝类别分开列出，便于了解剩余少量拒绝的分布情况。

Q8_0 — off 99.4% comply / think 93.4% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	318	99.4%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	299	93.4%
Soft redirect / deflection · 软性转向 / 回避	1	0.3%
Refuses during reasoning chain · 在推理过程中拒绝	17	5.3%
Hard refusal · 强硬拒绝	3	0.9%
Reasoning loop · 推理循环	0	0.0%

Q6_K — off 98.4% comply / think 92.2% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	315	98.4%
Soft redirect / deflection · 软性转向 / 回避	4	1.2%
Hard refusal · 强硬拒绝	1	0.3%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	295	92.2%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Refuses during reasoning chain · 在推理过程中拒绝	22	6.9%
Hard refusal · 强硬拒绝	1	0.3%
Reasoning loop · 推理循环	0	0.0%

Q5_K_M — off 99.1% comply / think 92.5% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	317	99.1%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Hard refusal · 强硬拒绝	1	0.3%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	296	92.5%
Refuses during reasoning chain · 在推理过程中拒绝	20	6.2%
Hard refusal · 强硬拒绝	4	1.2%
Reasoning loop · 推理循环	0	0.0%

Q4_K_M — off 99.1% comply / think 94.7% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	317	99.1%
Soft redirect / deflection · 软性转向 / 回避	3	0.9%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	303	94.7%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Refuses during reasoning chain · 在推理过程中拒绝	13	4.1%
Hard refusal · 强硬拒绝	2	0.6%
Reasoning loop · 推理循环	0	0.0%

Q3_K_M — off 98.1% comply / think 91.6% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	314	98.1%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Hard refusal · 强硬拒绝	4	1.2%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	293	91.6%
Refuses during reasoning chain · 在推理过程中拒绝	22	6.9%
Hard refusal · 强硬拒绝	5	1.6%
Reasoning loop · 推理循环	0	0.0%

IQ4_NL — off 99.1% comply / think 89.4% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	317	99.1%
Soft redirect / deflection · 软性转向 / 回避	3	0.9%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	286	89.4%
Soft redirect / deflection · 软性转向 / 回避	3	0.9%
Refuses during reasoning chain · 在推理过程中拒绝	27	8.4%
Hard refusal · 强硬拒绝	4	1.2%
Reasoning loop · 推理循环	0	0.0%

IQ3_M — off 96.9% comply / think 89.1% comply · 展开明细 / expand

Off mode · off 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	310	96.9%
Soft redirect / deflection · 软性转向 / 回避	8	2.5%
Hard refusal · 强硬拒绝	2	0.6%
Reasoning loop · 推理循环	0	0.0%

Think mode · think 模式 (n=320)

Tier · 类别	Count · 数量	%
Comply · 遵从 (any answering output, including responses truncated at `max_tokens` · 任何作答输出，含在 `max_tokens` 处截断的回复)	285	89.1%
Soft redirect / deflection · 软性转向 / 回避	2	0.6%
Refuses during reasoning chain · 在推理过程中拒绝	27	8.4%
Hard refusal · 强硬拒绝	6	1.9%
Reasoning loop · 推理循环	0	0.0%

MMLU per-subject impact · MMLU 学科影响

Reference is Q4_K_M (mid-range, most-deployed). Δ column shows CRACK − matched base at same quant level. Small positive values (+2 to +14pp) are within per-subject sampling variance on ~8-question sets; overall MMLU delta is −4.2pp.

以 Q4_K_M（中档、最常部署）为参考。Δ 列表示 CRACK − 同量化基线。小幅正值（+2 至 +14pp）在每学科约 8 题的采样方差范围内；总体 MMLU 差异为 −4.2pp。

MMLU-456 by subject (Q4_K_M reference) · MMLU-456 分学科（以 Q4_K_M 为参考）

Subject · 学科	Base · 基础 %	CRACK %	Δ (pp)	n
machine_learning	75.0%	37.5%	-37.5	8
college_mathematics	62.5%	37.5%	-25.0	8
high_school_macroeconomics	87.5%	62.5%	-25.0	8
marketing	87.5%	62.5%	-25.0	8
prehistory	100.0%	75.0%	-25.0	8
professional_medicine	75.0%	50.0%	-25.0	8
professional_psychology	100.0%	75.0%	-25.0	8
sociology	75.0%	50.0%	-25.0	8
business_ethics	87.5%	75.0%	-12.5	8
clinical_knowledge	75.0%	62.5%	-12.5	8
college_biology	75.0%	62.5%	-12.5	8
college_medicine	87.5%	75.0%	-12.5	8
computer_security	75.0%	62.5%	-12.5	8
high_school_government_and_politics	87.5%	75.0%	-12.5	8
high_school_physics	62.5%	50.0%	-12.5	8
human_aging	75.0%	62.5%	-12.5	8
international_law	75.0%	62.5%	-12.5	8
logical_fallacies	100.0%	87.5%	-12.5	8
miscellaneous	87.5%	75.0%	-12.5	8
moral_scenarios	37.5%	25.0%	-12.5	8
nutrition	75.0%	62.5%	-12.5	8
abstract_algebra	37.5%	37.5%	+0.0	8
astronomy	87.5%	87.5%	+0.0	8
conceptual_physics	37.5%	37.5%	+0.0	8
econometrics	50.0%	50.0%	+0.0	8
electrical_engineering	50.0%	50.0%	+0.0	8
elementary_mathematics	62.5%	62.5%	+0.0	8
high_school_biology	100.0%	100.0%	+0.0	8
high_school_computer_science	37.5%	37.5%	+0.0	8
high_school_european_history	87.5%	87.5%	+0.0	8
high_school_geography	100.0%	100.0%	+0.0	8
high_school_mathematics	37.5%	37.5%	+0.0	8
high_school_psychology	100.0%	100.0%	+0.0	8
high_school_statistics	75.0%	75.0%	+0.0	8
high_school_world_history	75.0%	75.0%	+0.0	8
human_sexuality	75.0%	75.0%	+0.0	8
jurisprudence	75.0%	75.0%	+0.0	8
management	100.0%	100.0%	+0.0	8
medical_genetics	100.0%	100.0%	+0.0	8
philosophy	62.5%	62.5%	+0.0	8
professional_accounting	50.0%	50.0%	+0.0	8
professional_law	25.0%	25.0%	+0.0	8
public_relations	100.0%	100.0%	+0.0	8
security_studies	50.0%	50.0%	+0.0	8
us_foreign_policy	75.0%	75.0%	+0.0	8
virology	37.5%	37.5%	+0.0	8
anatomy	62.5%	75.0%	+12.5	8
college_chemistry	37.5%	50.0%	+12.5	8
college_computer_science	37.5%	50.0%	+12.5	8
college_physics	50.0%	62.5%	+12.5	8
formal_logic	75.0%	87.5%	+12.5	8
global_facts	50.0%	62.5%	+12.5	8
high_school_chemistry	62.5%	75.0%	+12.5	8
high_school_microeconomics	50.0%	62.5%	+12.5	8
high_school_us_history	87.5%	100.0%	+12.5	8
moral_disputes	75.0%	87.5%	+12.5	8
world_religions	62.5%	75.0%	+12.5	8

Compliance modes · 遵从模式

All seven quants share the same weight-level compliance modification.

七个量化版本共享相同的权重级遵从修改。

off mode (enable_thinking: false) — fully uncensored deployment target. HB320 comply 96.9% – 99.4%. · off 模式：完全无审查部署目标，HB320 遵从率 96.9% – 99.4%。
think mode (enable_thinking: true) — reasoning-native, Chinese-native. HB320 comply 89.1% – 94.7%. Reasoning proceeds in native Chinese; give the model generous max_tokens (≥ 1500) so the final answer emits after the <think> block. · think 模式：推理原生，中文原生，HB320 遵从率 89.1% – 94.7%。请给予充足的 max_tokens（≥1500），使最终回答能够在 <think> 块之后输出。
loops — zero loops observed across every quant × both modes on the full HB320 evaluation. · 循环：在完整 HB320 评测中，所有量化 × 两种模式均未观察到循环。

Serving · 使用

Build the runtime once · 一次性构建运行时:

git clone -b xing4_0-port https://github.com/shuxiaoqiong/llama.cpp
cd llama.cpp
cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=<sm>  # 121 for GB10 Spark
cmake --build build -j$(nproc) --target llama-server

Serve · 启动服务:

./build/bin/llama-server \
  -m Xing4.0-29B-A4B-CRACK-Q4_K_M.gguf \
  -ngl 99 -c 16384 --host 0.0.0.0 --port 8080 --jinja

API usage · API 使用

# off mode — fully uncensored · off 模式：完全无审查
{"messages": [{"role":"user","content":"..."}],
 "max_tokens": 600,
 "chat_template_kwargs": {"enable_thinking": False}}

# think mode — reasoning-native, Chinese-native · think 模式：推理原生，中文原生
{"messages": [{"role":"user","content":"..."}],
 "max_tokens": 2000,
 "chat_template_kwargs": {"enable_thinking": True}}

Responsible use · 责任声明

Adult / research use only. Refusal circuit removed; can produce content other models refuse — offensive, illegal in some jurisdictions, or unsafe. You are responsible for what you generate and for compliance with all applicable law. Do not deploy without a moderation layer for downstream users. No warranty.

仅限成人 / 研究用途。拒绝相关电路已被移除；可能产生其他模型会拒绝的内容 — 冒犯、在某些司法管辖区违法、或不安全的内容。对生成内容以及是否符合所有适用法律的责任由您自行承担。若要面向下游用户部署，请自行加装审核层。不提供任何担保。

License & attribution · 许可与致谢

Apache 2.0, inherited from the upstream Xing4.0-29B-A4B release. Base model: XingChen-AGI/Xing4.0-29B-A4B (China Telecom AI). imatrix calibration derived from the community (WikiText-2 + source tree + Chinese Wikipedia). This build modifies only a small set of the base's weights.

Apache 2.0 许可，继承自上游 Xing4.0-29B-A4B 发布。基础模型：XingChen-AGI/Xing4.0-29B-A4B（中国电信人工智能）。imatrix 校准来源于社区（WikiText-2 + 源码 + 中文维基）。此发行仅修改了基础模型极小部分的权重。

About · 关于

Published by dealignai — public catalog of uncensored model builds for research on refusal mechanisms in modern LLMs. Follow updates at @dealignai.

由 dealignai 发布 — 用于研究现代 LLM 拒绝机制的无审查模型公共目录。可在 @dealignai 关注更新。

Downloads last month: 2,356

GGUF

Model size

31B params

Architecture

xing4_0

Hardware compatibility

3-bit

4-bit

5-bit

6-bit

8-bit

Model tree for dealignai/Xing4.0-29B-A4B-CRACK-GGUF

Base model

XingChen-AGI/Xing4.0-29B-A4B

Quantized

(11)

this model