---

license: other

license_name: nemotron-fable5-derivative

base_model: unsloth/NVIDIA-Nemotron-3-Super-120B-A12B

datasets:

• Glint-Research/Fable-5-traces

library_name: gguf

pipeline_tag: text-generation

tags:

• nemotron
• nemotron_h
• mamba2
• mixture-of-experts
• gguf
• llama.cpp
• lm-studio
• qlora
• reasoning
• agentic
• tool-use

---

Fabletron — Nemotron-3-Super-120B-A12B · Fable-5 (GGUF)

QLoRA fine-tune of NVIDIA Nemotron-3-Super-120B-A12B (120B-total / 12B-active hybrid

Mamba-2 + Latent-MoE, nemotron_h) on the pi_agent split of

Glint-Research/Fable-5-traces,

targeting reasoning, agentic planning, and tool-use. This repo holds the GGUF export for

LM Studio / llama.cpp.

Files

| File | Quant | Size |

|------|-------|------|

| Nemotron-3-Super-120B-fable5-Q4_K_M.gguf | Q4_K_M | ≈80.1 GiB (86.05 GB) |

> Requires a llama.cpp build with nemotron_h_moe support (PR #18058 or later). Recent

> LM Studio runtimes include it. Uses the ChatML template baked into the GGUF.

Training

• Method: QLoRA (4-bit NF4 base) via Unsloth. LoRA r=8, α=16, dropout=0, bias=none.

Targets (by regex) attention q/k/v/o_proj, Mamba-2 in_proj/out_proj, and the

up/down_proj of all 512 routed experts + the shared expert — this MoE is non-gated,

so up/down_proj is the full expert FFN (there is no gate_proj). The MoE router

(mixer.gate, an nn.Parameter) and the MoE-latent fc1/fc2_latent_proj stay frozen.

• Data: Glint-Research/Fable-5-traces, config pi_agent, ChatML, response-only loss.
• Schedule: 1 epoch (248 steps), grad-accum 16, lr 2e-5 cosine + warmup.
• Final training loss: ≈0.74 true per-token CE (final-phase mean; Unsloth logs loss ×

grad-accum, so the raw logged value is ÷16 here).

• Evaluation (held-out Fable-5 pi_agent, 81 rows, base → fine-tuned): cross-entropy

1.12 → 0.78, perplexity 3.07 → 2.18.

• Date: 2026-06-29.
• LoRA adapter (pre-merge) + other formats: greghavens/fabletron-nemotron-3-super-120b.

Provenance & license

This is a derivative of two upstream works, and downstream use must comply with both:

• Base model: NVIDIA Nemotron-3-Super-120B-A12B — see NVIDIA's model license.
• Dataset: Glint-Research/Fable-5-traces — AGPL-3.0, and distilled from Anthropic

Claude outputs (subject to Anthropic's usage terms).

Released as a research artifact. Review the upstream licenses before redistribution or

production use. license: other reflects the combination of the above, not a single license.

Intended use

Research on hybrid Mamba-2/MoE reasoning models, agentic/tool-use experimentation. Not

safety-aligned beyond the base model; evaluate before any production deployment.

<!-- FABLETRON_BENCHMARKS_START -->

Benchmarks

Fabletron is a Fable-5 fine-tune of Nemotron-3-Super-120B-A12B. This repository hosts the GGUF (Q4\_K\_M — for llama.cpp / LM Studio / Ollama) build. The results below were produced through a deployed NeMo Evaluator at full context length (262 144) with a 32 768-token generation budget, comparing base Nemotron-3-Super against the Fabletron fine-tune in its two served formats — GGUF (llama.cpp) and NVFP4 (vLLM / Blackwell).

Accuracy

!Fabletron benchmark scores

| Benchmark | Base Nemotron 3 Super | Fabletron (GGUF) | Fabletron (NVFP4) |

|---|---|---|---|

| IFEval | 89.6 | 90.4 | 87.4 |

| GSM8K | 96.6 | 96.5 | 96.5 |

| MMLU-Pro | 81.1 | 81.5 | 81.2 |

| HumanEval+ | 87.8 | 89.6 | 89.6 |

| MBPP+ | 95.5 | 96.6 | 96.8 |

| BFCL (v3 · multi_turn + ast) | 31.7 | 53.6 | 53.9 |

Scores are % (higher is better). The standout is function-calling (BFCL): 31.7 → 53.6 — base Nemotron largely fails multi-turn and parallel tool use, which the fine-tune resolves (single-tool calls 0.8 → 85.7, multi-turn 0 → 45). General reasoning and coding hold or improve slightly, and NVFP4 tracks the GGUF fine-tune within noise — the only give-back to quantization is IFEval strict (≈2 points).

Throughput

!Fabletron throughput

Mean per-request decode rate across the 6 benchmark families (tok/s, client-side, parallelism 4): base 45.7 · Fabletron GGUF 46.3 · Fabletron NVFP4 63.8. The NVFP4 build serves at ≈1.4× the GGUF (llama.cpp) rate on an RTX PRO 6000 (Blackwell).

_{BFCL = Berkeley Function-Calling Leaderboard v3 (multi_turn,ast, native tool-calling, via nvidia_bfcl 26.3). Base and GGUF share the llama.cpp engine; NVFP4 runs on vLLM. Output truncation at this budget is ≈0 across families (MMLU-Pro shows ≈3.5% length-capped generations, intrinsic to the benchmark and near-identical for all three models, so it does not bias the comparison).}

<!-- FABLETRON_BENCHMARKS_END -->