---

license: apache-2.0

base_model:

• MuXodious/GLM-4.7-Flash-absolute-heresy

library_name: ik_llama.cpp

pipeline_tag: text-generation

tags:

• gguf
• quantized
• ik_llama.cpp
• roleplay
• natural language
• english
• trellis
• imatrix
• mixed precision

quantized_by: Koshkasa

base_model_relation: quantized

datasets:

• eaddario/imatrix-calibration

---

MODEL CARD INCOMPLETE

BENCHMARKS PENDING

!!! quantized for use with ik_llama.cpp and its derivatives !!!

!!! incompatible with mainline llama.cpp as of commit #34af94c !!!

What's that?

An experimental, English-roleplay-oriented, ik_llama.cpp-only MuXodious/GLM-4.7-Flash-absolute-heresy quantization for users who want an IQ3_M-sized file but prefer to spend precision on MLA, routing, shared experts, and output-sensitive tensors.

Rationale

I wanted a quant that would fit my GPU with some context and minimal offload. Simple as.

This gguf was made to compete with IQ3_M/IQ4_XS by compressing sparse expert ffn tensors in SOTA low-bit quant types (IQ3_KT/IQ4_KT) while protecting the most compression-sensitive, architecture critical tensors. I had concerns with mainline quant method compromises - such as shared experts in 3-bit, and the MLA KV "condensation" weights (attn_kv_a_mqa) being quanted lower than attn_k_b (the layer extracting Keys from the shared latent attention vector). Fearing that overcompressed MLA latent projection would mangle reconstructed attention states, I went for a much higher quantization for these. The recipe provided is, however, an exploratory MLA/MoE allocation, not gospel.

The proportion of ffn_exps parameters in the entire model is 92.46%. And 2.12% for lm_head and embeddings. Meaning EVERYTHING else - the shared exps, the attention tensors - is 3 GB in bf16. These also happen to be quantization sensitive tensors. As a prime example, keeping expert routing weights in bf16 across the entire gguf has cost... 7.5 MB over IQ3_S.

The imatrix was generated on wrapped natural language english text from eaddario/imatrix-calibration, using kld-sweep-dataset by cmhamiche. The imatrix was not calibrated for STEM, mathematics, code, or non-English languages. I was building it for my use purposes.

However, I'm not opposed to making a trellis quant for other use cases if anyone needs it.

UPD: since the IQ3_KT ffn_down_exps also seems to be perfectly functional, it's provided as well.

Mixed Trellis

| Component / Role | Tensor (Layer) | Dense Block 0 (1x) | MoE/MLA Blocks (46x) |

| :--- | :--- | :---: | :---: |

| Global Layers | token_embd.weight <br> output.weight | iq5_k <br> — | — <br> iq6_k |

| MLA Compressed | attn_q_a.weight <br> attn_kv_a_mqa.weight | bf16 <br> bf16 | bf16 <br> bf16 |

| MLA Attention | attn_q_b <br> attn_k_b / attn_v_b <br> attn_output.weight | q8_0 <br> q8_0 <br> q8_0 | q8_0 <br> q8_0 <br> iq6_k |

| MoE Routing | ffn_gate_inp.weight <br> exp_probs_b | — <br> — | bf16 <br> f32 |

| MoE Experts | ffn_down_exps.weight <br> ffn_gate_exps.weight <br> ffn_up_exps.weight | — <br> — <br> — | iq4_kt <br> iq3_kt <br> iq3_kt |

| Shared Experts | ffn_down_shexp <br> ffn_gate_shexp <br> ffn_up_shexp | — <br> — <br> — | q8_0 <br> q8_0 <br> q8_0 |

| Dense MLP | ffn_down / ffn_gate / ffn_up | q8_0 | — |

IQ3_KT

| Component / Role | Tensor (Layer) | Dense Block 0 (1x) | MoE/MLA Blocks (46x) |

| :--- | :--- | :---: | :---: |

| Global Layers | token_embd.weight <br> output.weight | iq5_k <br> — | — <br> iq6_k |

| MLA Compressed | attn_q_a.weight <br> attn_kv_a_mqa.weight | bf16 <br> bf16 | bf16 <br> bf16 |

| MLA Attention | attn_q_b <br> attn_k_b / attn_v_b <br> attn_output.weight | q8_0 <br> q8_0 <br> q8_0 | q8_0 <br> q8_0 <br> iq6_k |

| MoE Routing | ffn_gate_inp.weight <br> exp_probs_b | — <br> — | bf16 <br> f32 |

| MoE Experts | ffn_down_exps.weight <br> ffn_gate_exps.weight <br> ffn_up_exps.weight | — <br> — <br> — | iq3_kt <br> iq3_kt <br> iq3_kt |

| Shared Experts | ffn_down_shexp <br> ffn_gate_shexp <br> ffn_up_shexp | — <br> — <br> — | q8_0 <br> q8_0 <br> q8_0 |

Cheers

Z.ai - the base model.

ikawrakow and contributors of ik_llama.cpp - I probably misused your creation.

MuXodious - for letting the model swear.

cmhamiche - for accessible, ready-to-use dataset construction tool.

eaddario - for the imatrix dataset.