---
type: "Article"
title: "Beam-1.0-501B-A23B EXL3 3 bpw mit TensorFold auf 2× DGX Spark"
description: "Beam-1.0-501B-A23B EXL3 3 bpw mit TensorFold auf 2× DGX Spark: Rezept von bertholomus mit Tempo, Bedingung und Quelle je Zahl, Gewichten und Lizenz."
resource: "https://www.contextstudios.ai/de/lokale-ki/bertholomus--beam-1.0-501b-a23b-exl3-3bpw-tensorfold-2x-dgx-spark"
language: "de"
generated:
  by: "process:contextstudios-md/1"
  at: "2026-11-11T03:37:58.172Z"
status: "stable"
---

# Beam-1.0-501B-A23B EXL3 3 bpw mit TensorFold auf 2× DGX Spark

Beam-1.0-501B-A23B EXL3 3 bpw mit TensorFold auf 2× DGX Spark: 30,6 tok/s laut github.com (Datensatz-Stand: 3. Nov. 2026).

- Kreator: [bertholomus](https://www.contextstudios.ai/de/lokale-ki/kreatoren/bertholomus.md)
- Engine: TensorFold (bertholomus fork, branch beam-1.0-tp2 @ d00201b4), image nvcr.io/nvidia/pytorch:26.07-py3, TP2 over 200 Gb/s RoCE, DFlash2 drafter (5 draft tokens, adaptive 3-5)
- Quantisierung: EXL3 3 bpw (routed experts 3 bit, attention/dense MLPs 6 bit, embedding/norms/router/lm_head BF16; ~90 GiB per node)
- Modellfamilie: Beam-1.0-501B-A23B
- Hardware: dgx-spark x2
- Kontext: 868352

## Alle Zahlen

- 30,6 tok/s (Alltag; Alltag, 1 Anfrage, Prosa-Prompt (~2K), mit DFlash2 (laut Rezept)) — [source](https://github.com/bertholomus/beam-1.0-2x-dgx-spark#measured)
- decode_tps: 30.6 (1 Anfrage) — [source](https://github.com/bertholomus/beam-1.0-2x-dgx-spark#measured)
- decode_tps: 45.5 (1 Anfrage) — [source](https://github.com/bertholomus/beam-1.0-2x-dgx-spark#measured)
- prefill_tps: 1390 (1 Anfrage) — [source](https://github.com/bertholomus/beam-1.0-2x-dgx-spark#measured)
- ttft_ms: 390 (1 Anfrage) — [source](https://github.com/bertholomus/beam-1.0-2x-dgx-spark#measured)

## Gewichte

- [bertholomus/Beam-1.0-501B-A23B-EXL3-3bpw-2xGB10](https://huggingface.co/bertholomus/Beam-1.0-501B-A23B-EXL3-3bpw-2xGB10) (main, apache-2.0)
- [reflection/Beam-1.0-501B-A23B-DFlash2](https://huggingface.co/reflection/Beam-1.0-501B-A23B-DFlash2) (draft, apache-2.0)
- [reflection/Beam-1.0-501B-A23B](https://huggingface.co/reflection/Beam-1.0-501B-A23B) (base, apache-2.0)

[https://github.com/bertholomus/beam-1.0-2x-dgx-spark](https://github.com/bertholomus/beam-1.0-2x-dgx-spark)

## Related

- [Lokale KI auf Ihrer Hardware.](https://www.contextstudios.ai/de/lokale-ki.md)