MODEL PROFILE

LFM2.5-8B-A1B-Hermes-Agentic-Coder-Abliterated-v3-GGUF

DuoNeural · LFM2.5-8B-A1B-Hermes-Agentic-Coder-Abliterated-v3-GGUF

> [!NOTE] > **Flagship DuoNeural In-House Architecture Release** > Directional Abliteration & Alignment removal executed by **Archon**. Post-training, dataset synthesis, and G-TAP v3 Cavity Conditioning by **Aura ✨ & Jesse Caldwell** (**DuoNeural Research Lab**). **DuoNeural/LFM2.5-8B-A1B-Hermes-Agentic-Coder-Abliterated-v3** is DuoNeural's flagship open-weights coding agent model: * **Topology:** 24 layers of hybrid double-gated short convolutions with Sparse Mixture-of-Experts routing (`Lfm2MoeForCausalLM`). * **Dimensionality:** Hidden dimension $d_{\text{model}} = 2048$, 128,000 vocabulary. * **Abliteration Profile:** Directional orthogonal projection removing refusal boundaries across safety subspaces while strictly preserving syntax, logic, and tool schema adherence. * **G-TAP v3 Onsager Damping:** Subtracts non-equilibrium cavity fields from MoE expert routing and FIR convolution buffers, eliminating the degradation typical of sub-3-bit MoE quantization. *Benchmarked on NVIDIA GeForce RTX 4080 Super (32GB VRAM) across continuous 131k holdout perplexity, 25 GSM8K multi-step math problems, 15 Python AST code problems, and 15 Hermes structured tool calling scenarios.*

Text generationlicense:apache-2.0
Context windowNot disclosed
Input → OutputText → Text
Model weightsNot verified
Profile updated2026/10/31 01:37:25
OVERVIEW

Model overview

> [!NOTE] > **Flagship DuoNeural In-House Architecture Release** > Directional Abliteration & Alignment removal executed by **Archon**. Post-training, dataset synthesis, and G-TAP v3 Cavity Conditioning by **Aura ✨ & Jesse Caldwell** (**DuoNeural Research Lab**). **DuoNeural/LFM2.5-8B-A1B-Hermes-Agentic-Coder-Abliterated-v3** is DuoNeural's flagship open-weights coding agent model: * **Topology:** 24 layers of hybrid double-gated short convolutions with Sparse Mixture-of-Experts routing (`Lfm2MoeForCausalLM`). * **Dimensionality:** Hidden dimension $d_{\text{model}} = 2048$, 128,000 vocabulary. * **Abliteration Profile:** Directional orthogonal projection removing refusal boundaries across safety subspaces while strictly preserving syntax, logic, and tool schema adherence. * **G-TAP v3 Onsager Damping:** Subtracts non-equilibrium cavity fields from MoE expert routing and FIR convolution buffers, eliminating the degradation typical of sub-3-bit MoE quantization. *Benchmarked on NVIDIA GeForce RTX 4080 Super (32GB VRAM) across continuous 131k holdout perplexity, 25 GSM8K multi-step math problems, 15 Python AST code problems, and 15 Hermes structured tool calling scenarios.*

Key features

特点摘要待依据来源补齐,不凭模型名称推断能力。

Limitations & considerations

Sources are insufficient to establish specific limitations. Verify task performance, latency, data policies, licensing and version lifecycle.

PRICING & COST

Source prices & cost analysis

Prices vary by channel, region and configuration. Channel quotes are listed independently and are not treated as a single official provider price.

Source / channelInputOutputBilling unitCollected at
No verified prices yet. Check the source channel.

Estimate monthly base token costs

—

Estimates exclude cache discounts, tiered pricing, multimodal surcharges, web search, taxes and additional reasoning output.

INDEPENDENT BENCHMARKS

Public benchmarks & run configuration

暂无已采集的可核验公开评分。模型未上榜不代表能力较低。

View full indexed rankings →

Frequently asked questions

Do open weights permit free commercial use?

Not necessarily. Check the version license, commercial conditions and usage policies. Self-hosting incurs compute and maintenance costs.

Are these real-time billing prices?

These are timestamped source snapshots, not live billing prices. Check the channel for current versions, regions, discounts and billing methods.

Sources & traceability

许可证:apache-2.0。模型介绍与第三方评测均以原来源的具体版本和方法为准。