RESMP-DEV/LFM2.5-Encoder-350M-Code-MXFP8-GPTQ
Feature Extraction • 0.1B • Updated • 69
AI & ML interests
Bleeding-Edge AI R&D.
Recent Activity
View all activity
Organization Card
■ INDEPENDENT APPLIED RESEARCH
RESMP / COMPOSITION LAB
ARTIFACT
SYSTEMS
SYSTEMS
FIELD LAB
REAL HW
ACTIVE
Evidence-first
model systems.
We work in the seams between models, quantizers, runtimes, datasets, agents, and hardware—combining strong artifacts into systems their individual parts did not imply.
NOT A MODEL FACTORY / NOT A FINETUNE MILL / NOT A SINGLE-DISCIPLINE LAB
PRECISION
NVFP4, FP8, MXFP4/8,
GPTQ, mixed precision
SYSTEMS
CUDA, Apple Silicon,
native inference paths
EVIDENCE
Controls, hashes,
manifests, limitations
■ BENCHMARK INSTRUMENT
One visual system.
Any evaluation matrix.
Data-driven small multiples with fixed series order, direct labels, negative-score support, SVG export, and print output.
OPEN CHART SYSTEM ↗HOW WE PUBLISH / MEASURED · REPRODUCIBLE · HONEST
The artifact is only half the release.
The open ecosystem supplies extraordinary parts. Our work is composition: finding unexpected fit, proving the system, and publishing its boundary conditions—pinned revisions, licenses, controls, runtime constraints, evaluations, receipts, and negative results.
01
Artifact composition
Models, methods, data, and runtimes combined into systems with new capabilities.
02
Code intelligence
Multilingual retrieval encoders, calibrated quantization, and local serving.
03
Agentic systems
Reasoning traces, continuous distillation, mutation, and adversarial workloads.
04
Accessible computing
Speech and language systems built around accessibility-centered tasks.
models 18
RESMP-DEV/LFM2.5-Encoder-350M-Code-MXFP4-GPTQ
Feature Extraction • 0.1B • Updated • 124 • 1
RESMP-DEV/LFM2.5-Encoder-350M-Code-BF16
Feature Extraction • 0.4B • Updated • 25
RESMP-DEV/LFM2.5-Encoder-230M-Code-MXFP8-GPTQ
Feature Extraction • 0.1B • Updated • 32
RESMP-DEV/LFM2.5-Encoder-230M-Code-MXFP4-GPTQ
Feature Extraction • 92.6M • Updated • 32
RESMP-DEV/LFM2.5-Encoder-230M-Code-BF16
Feature Extraction • 0.2B • Updated • 24
RESMP-DEV/jina-code-embeddings-1.5b-mxfp4-block-gptq
Feature Extraction • 0.4B • Updated • 149
RESMP-DEV/MiMo-V2.5-ASR-FP8
Automatic Speech Recognition • 8B • Updated • 13
RESMP-DEV/Qwen3.5-27B-GPTQ-Hybrid
27B • Updated • 11
RESMP-DEV/Qwen3.5-35B-A3B-GPTQ-Hybrid
36B • Updated • 7