Boldt-DC-1B German IT 16K — DPO + SLERP refinement

A SLERP-merged variant of mayflowergmbh/boldt-dc-1b-german-it-16k. Built by merging the SFT model with a DPO-tuned checkpoint of itself, following the same-model-merging approach described in the LFM2 technical report §4.4 (Liquid AI, arXiv:2511.23404). Same architecture, same context length, same chat format as the SFT release.

Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "mayflowergmbh/boldt-dc-1b-german-it-16k-dpo"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, dtype=torch.bfloat16, device_map="cuda")

messages = [{"role": "user", "content": "Erkläre kurz, was eine Funktion in Python ist."}]
prompt = tok.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)
inputs = tok(prompt, return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=200, do_sample=False)
print(tok.decode(out[0, inputs.input_ids.shape[-1]:], skip_special_tokens=False))

generation_config.eos_token_id = [0, 32003] covers both <|endoftext|> and <|end|>.

Recipe

1. SFT model (mayflowergmbh/boldt-dc-1b-german-it-16k): plain transformers+peft SFT of Boldt/Boldt-DC-1B, 7000 steps at 16K context.
2. DPO checkpoint: TRL DPOTrainer, loss_type="sigmoid", β=0.3, rpo_alpha=0.5 (NLL anchor on chosen — prevents the "response suppression" failure mode documented in 3D-Properties of DPO, Yan et al. 2024, arXiv:2406.07327). LR 5e-7, 800 steps, LoRA r=32 on QKV/MLP. Dataset: mayflowergmbh/boldt-dc-1b-orpo-onpolicy-de length-filtered to |chosen|/|rejected| ≤ 3 (54k → 22k pairs).
3. SLERP merge via mergekit at t=0.5, dtype=bfloat16, tokenizer_source: union. ~30 seconds on a single A6000.

Why merge: the SFT model preserves more reasoning capacity (commonsense benchmarks regress less than under pure DPO), while the DPO model has slightly better chat-style behaviour. SLERP at t=0.5 recovers both. The LFM2 paper documents the same observation for full-model merging at the 1.2B scale.

Evaluation (lm-evaluation-harness, German tier 1)

The merge is the highest-mean working variant in the SFT/DPO/merge family — +0.25 pp over pure SFT, +0.11 pp over pure DPO. Largest individual gains: arc_de recovers +0.34 pp from the DPO regression, belebele_deu_Latn adds +0.89 pp over SFT. Per-task deltas are within stderr (~±1.5 pp), but the direction is consistent across tasks and the result reproduces the LFM2 paper's claim that same-model merging recovers task-specific knowledge that preference tuning erodes.

No public 1B-class German chat model published Q4 2025 – Q1 2026 has been found to meaningfully exceed the Boldt/Boldt-DC-1B base on these tier-1 averages without teacher-model distillation. This release does not close that gap, but it is the highest tier-1 mean among models in this family that also generate coherent German.

Mergekit config

slices:
  - sources:
      - model: mayflowergmbh/boldt-dc-1b-german-it-16k        # SFT
        layer_range: [0, 16]
      - model: <DPO-tuned checkpoint of the SFT>
        layer_range: [0, 16]
merge_method: slerp
base_model: mayflowergmbh/boldt-dc-1b-german-it-16k
parameters:
  t: 0.5
dtype: bfloat16
tokenizer_source: union

Known limitations

Inherits all of the SFT base's limits — math arithmetic is unreliable (1.25 B ceiling), factual recall has typical small-model errors, no tool-use / function-calling training, long-context use beyond ~8 K is untested.

License

Apache-2.0 (inherits from the base model).