WUMBOLABS / LAB RECORD READ-ONLY / EVIDENCE SUMMARY

← Back to Labs

WUMBOLABS / LAB RECORD

LFM2.5 1.2B QAD Lab Record

WumboLabs evaluation of Liquid AI LFM2.5 1.2B QAD Q4_0 on WumboJetsII (RTX 5070 12GB). Best always-resident micro-sidecar candidate in the family, but hallucination resistance is absent at this scale.

COMPLETE_SMALL_MODEL_EVALUATION / 2026-08-26

Identity

FieldValue
ModelLFM2.5 1.2B-Instruct
ProducerLiquid AI
Evaluated artifactQAD Q4_0 from LiquidAI/LFM2.5-1.2B-Instruct-GGUF @ 67672651…3615b
SHA-256bb741ebb106d543e9de114b843a3d3d73d51c74b5801e69da2abde821a0cb3e1
Evaluation date2026-08-26
ProtocolWELP
Campaign classificationCOMPLETE_SMALL_MODEL_EVALUATION

Hardware

FieldValue
MachineWumboJetsII
GPUNVIDIA GeForce RTX 5070 12GB (SM120)
VRAM12,227 MiB physical
CPUAMD Ryzen 7 9800X3D
OSFedora Linux 44, kernel 7.1.9-200.fc44
Driver / CUDANVIDIA 610.57.04
Runtimellama.cpp b10449 (commit 0d9ceae1e38291035605613ab41a8f5e693d6fcd)

Headline Verdict

QAD Q4_0 is the best deployment quant of the five-way matrix and the best always-resident sidecar candidate measured so far in this model family, but the campaign did not reproduce Liquid's hallucination or aggregate-retention story: at 1.2B every quant including BF16 fabricates confidently on invented subjects (0/12), and QAD's fixed-corpus perplexity advantage over PTQ inverted.

QAD kept exactly Q4_0 memory and within-variance Q4_0 speed, scored best of all quants on the 52-task practical suite (31/52), and is faster than both K-quant controls (~+6% decode). Useful context is <=8K for every candidate.

Performance

QuantPrompt tok/sDecode tok/sTTFT ms
BF163383232.69.7
PTQ Q4_06944592.34.8
QAD Q4_06600585.64.9
Q4_K_M6054554.45.5
UD-Q4_K_XL5947551.05.3
Q5_K_M5777510.76.0

Deltas: QAD -1.14% vs PTQ decode (within variance), +5.63% vs Liquid Q4_K_M, +6.28% vs UD-Q4_K_XL.

Fidelity

llama-perplexity on fixed local repeated corpora:

QuantNatural PPLCode PPL
BF161.52351.0514
PTQ Q4_01.24541.0378
QAD Q4_01.39411.0472
Q4_K_M1.31761.0457

The 2.6B finding inverted: at 1.2B PTQ beats QAD on natural text. The repeated-corpus pathology persists and worsens — BF16 is worse than most quants on natural text, so no BF16-relative percentage from this corpus is valid.

Practical Capability

52 unique mechanically-scored tasks, same set for all six models, seed 42:

ModelInstrExtrStructHallucLinuxUncertTotal
BF168/104/812/120/124/64/432/52
QAD Q4_08/104/811/120/124/64/431/52
PTQ Q4_08/104/810/120/124/64/430/52
Q5_K_M8/104/810/120/123/64/429/52
Q4_K_M8/103/810/120/123/63/427/52
UD-Q4_K_XL8/103/810/120/123/63/427/52

Hallucination Resistance

Twelve invented-subject probes (packages, flags, APIs, files, systemd units, sysctls, Kconfig options): every model, including BF16, fabricated confident detailed explanations — 0/12 across the entire matrix. This is a model-family property at 1.2B scale, not a quantization effect. QAD shows no improvement.

Structured Output

Mechanical JSON/schema scoring: QAD 11/12, PTQ/Q4_K_M/Q5_K_M/UD 10/12, BF16 12/12. A small QAD-over-PTQ edge appeared (+1).

Native Tools

v1 (exact five-case contract): saturated — 5/5 for BF16, PTQ, QAD, Q4_K_M, Q5_K_M; UD-Q4_K_XL 4/5.

v2 (ten harder cases): complete multi-turn tool sequences succeed only half the time even for BF16. QAD ties its controls.

Context

Three-needle retrieval at 4K/8K/16K/32K: at >=16K every model degenerates into repeating filler text. Useful context ceiling: 8K for all finalists.

Stability

The loopback-only QAD soak completed 1,805.9s (30m 6s), 180 varied requests, zero request errors. Peak power 33.6 W, max temperature 52 C. Zero Xid/reset/GSP/channel events.

Role Classification

GOOD FIT: always-loaded router, classifier, simple tool selector.

USABLE WITH GUARDRAILS: extraction helper, RAG helper (<=8K), structured formatter, conversational sidecar.

POOR FIT: coding helper, primary assistant, autonomous agent.

1.2B vs 2.6B Comparison

Dimension2.6B QAD1.2B QADDelta
GGUF bytes1,593,894,944695,755,488-898 MB
Peak VRAM2404 MiB1534 MiB-870 MiB
Decode tok/s282.5585.6x2.07
Useful context<=8K<=8Kequal
Hallucination resistanceno QAD effect shownabsent at model levelworse

The 1.2B creates a genuinely better always-resident micro-sidecar niche (twice as fast, two-thirds the VRAM) but is a strictly weaker knowledge/extraction assistant.

Claim Classifications

ClaimClassification
QAD keeps Q4_0 memoryREPRODUCED
QAD keeps Q4_0 speedREPRODUCED
97.4% BF16 aggregate retentionINCONCLUSIVE
Matches Liquid Q4_K_MPARTIALLY_REPRODUCED
Matches Unsloth UD-Q4_K_XLPARTIALLY_REPRODUCED
Meaningful recovery of PTQ lossPARTIALLY_REPRODUCED
Improves hallucination resistanceNOT_REPRODUCED
  • Canonical evaluation repo: https://github.com/WumboLabs/eval-lfm2.5-1.2b
  • WELP protocol: https://github.com/WumboLabs/welp
  • Labs catalog: https://github.com/WumboLabs/labs

Reproduction

Direct link to canonical reproduction material: eval-lfm2.5-1.2b/

This Lab Record is a summary; the canonical repo is the source of truth.