Spaetzle-v31-7b

Spaetzle-v31-7b is a merge of the following models using LazyMergekit:

• yleo/EmertonMonarch-7B
• cstr/spaetzle-v8-7b

Model	AGIEval	GPT4All	TruthfulQA	Bigbench	Average
Spaetzle-v31-7b	46.23	76.6	69.58	46.79	59.8

AGIEval

Task	Version	Metric	Value		Stderr
agieval_aqua_rat	0	acc	28.74	±	2.85
		acc_norm	27.56	±	2.81
agieval_logiqa_en	0	acc	39.63	±	1.92
		acc_norm	40.25	±	1.92
agieval_lsat_ar	0	acc	24.35	±	2.84
		acc_norm	24.35	±	2.84
agieval_lsat_lr	0	acc	54.31	±	2.21
		acc_norm	54.12	±	2.21
agieval_lsat_rc	0	acc	65.80	±	2.90
		acc_norm	66.54	±	2.88
agieval_sat_en	0	acc	79.13	±	2.84
		acc_norm	79.61	±	2.81
agieval_sat_en_without_passage	0	acc	46.12	±	3.48
		acc_norm	45.15	±	3.48
agieval_sat_math	0	acc	35.00	±	3.22
		acc_norm	32.27	±	3.16

Average: 46.23%

GPT4All

Task	Version	Metric	Value		Stderr
arc_challenge	0	acc	64.76	±	1.40
		acc_norm	66.89	±	1.38
arc_easy	0	acc	86.66	±	0.70
		acc_norm	82.83	±	0.77
boolq	1	acc	87.80	±	0.57
hellaswag	0	acc	67.43	±	0.47
		acc_norm	85.85	±	0.35
openbookqa	0	acc	38.00	±	2.17
		acc_norm	48.80	±	2.24
piqa	0	acc	83.57	±	0.86
		acc_norm	84.71	±	0.84
winogrande	0	acc	79.32	±	1.14

Average: 76.6%

TruthfulQA

Task	Version	Metric	Value		Stderr
truthfulqa_mc	1	mc1	53.37	±	1.75
		mc2	69.58	±	1.48

Average: 69.58%

Bigbench

Task	Version	Metric	Value		Stderr
bigbench_causal_judgement	0	multiple_choice_grade	56.84	±	3.60
bigbench_date_understanding	0	multiple_choice_grade	66.94	±	2.45
bigbench_disambiguation_qa	0	multiple_choice_grade	44.57	±	3.10
bigbench_geometric_shapes	0	multiple_choice_grade	21.17	±	2.16
		exact_str_match	0.28	±	0.28
bigbench_logical_deduction_five_objects	0	multiple_choice_grade	31.80	±	2.08
bigbench_logical_deduction_seven_objects	0	multiple_choice_grade	22.57	±	1.58
bigbench_logical_deduction_three_objects	0	multiple_choice_grade	56.00	±	2.87
bigbench_movie_recommendation	0	multiple_choice_grade	45.40	±	2.23
bigbench_navigate	0	multiple_choice_grade	52.80	±	1.58
bigbench_reasoning_about_colored_objects	0	multiple_choice_grade	70.65	±	1.02
bigbench_ruin_names	0	multiple_choice_grade	50.67	±	2.36
bigbench_salient_translation_error_detection	0	multiple_choice_grade	30.66	±	1.46
bigbench_snarks	0	multiple_choice_grade	71.27	±	3.37
bigbench_sports_understanding	0	multiple_choice_grade	74.34	±	1.39
bigbench_temporal_sequences	0	multiple_choice_grade	49.80	±	1.58
bigbench_tracking_shuffled_objects_five_objects	0	multiple_choice_grade	22.16	±	1.18
bigbench_tracking_shuffled_objects_seven_objects	0	multiple_choice_grade	18.57	±	0.93
bigbench_tracking_shuffled_objects_three_objects	0	multiple_choice_grade	56.00	±	2.87

Average: 46.79%

Average score: 59.8%

Elapsed time: 02:09:50

🧩 Configuration

models:
  - model: cstr/spaetzle-v8-7b
    # no parameters necessary for base model
  - model: yleo/EmertonMonarch-7B
    parameters:
      density: 0.60
      weight: 0.3
merge_method: dare_ties
base_model: cstr/spaetzle-v8-7b
parameters:
  int8_mask: true
dtype: bfloat16
random_seed: 0
tokenizer_source: base

💻 Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "cstr/Spaetzle-v31-7b"
messages = [{"role": "user", "content": "What is a large language model?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])

EU AI Act Art. 53 — provider obligations

Added 2026-08-02 during an account-wide provenance review.

This is a model merge, not a format conversion. Most cstr/* repositories are GGUF conversions, where the upstream research team remains the provider of the model and the conversion changes only the numeric representation of the weights. A merge produces a model that did not previously exist, so under Regulation (EU) 2024/1689 the maintainer of this repository is plausibly the provider of it, and the duties that survive the Art. 53(2) free-and-open-source exemption — Art. 53(1)(c) and 53(1)(d) — attach here rather than upstream.

Art. 53(1)(c) — copyright policy. No training corpus was assembled by this repository. Merging combines weights that other providers already published; it performs no text or data mining, so no rights reservation under Art. 4(3) of Directive (EU) 2019/790 was engaged by this step. Copyright questions arising from how the constituent models were themselves trained attach to their respective providers. Any credible claim that this repository redistributes material it has no right to redistribute will be acted on — contact via the Community tab.

Art. 53(1)(d) — training content. No data was used to train this model: it is a weight-space combination of models trained by others, and its training content is theirs. All 1 constituent models this card names are still published, so the chain can be followed from here.

Licence — resolved 2026-08-02. cc-by-nc-4.0. yleo/EmertonMonarch-7B is CC-BY-NC-4.0. The NonCommercial term propagates to anything containing those weights.

This was derived from the mergekit configuration reproduced in this card by resolving each named constituent's licence on the Hub and taking the most restrictive, rather than assumed from the model family. An earlier revision of this section said the terms were unresolved; they are resolved now, and the method is recorded so the conclusion can be checked rather than trusted.