DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data
Paper • 2608.13517 • Published • 32
DFM-Mimir AWQ FP4
This is the 4-bit GPTQ (W4A16_ASYM) quantized version of DFM-Mimir — a 1-billion-parameter HRM-Text language model trained from scratch delivering competitive performance for English and state-of-the-art performance for Danish using only permissible post-training data.
Quantization Details
| Property | Value |
|---|---|
| Method | GPTQ (W4A16_ASYM — 4-bit int asymmetric weights, 16-bit activations) |
| Library | llmcompressor 0.13.0 (GPTQModifier) |
| Weight format | pack-quantized (int4) |
| Group size | 128 (per-group quantization) |
| Actorder | static |
| Ignored | lm_head (kept in bf16) |
| Calibration | 256 samples from English benchmarks (MMLU 128, BoolQ 32, HellaSwag 32, GSM8K 32, Winogrande 32) |
| Max seq len | 2048 (calibration) |
| Model size | 2.12 GB (vs ~3.2 GB bf16) |
Loading
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(
"schneiderkamplab/DFM-Mimir-AWQ-FP4",
torch_dtype="bfloat16",
device_map="auto",
trust_remote_code=True,
)
tokenizer = AutoTokenizer.from_pretrained("schneiderkamplab/DFM-Mimir-AWQ-FP4")
messages = [
{"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
messages,
add_generation_prompt=True,
tokenize=True,
return_dict=True,
return_tensors="pt",
).to(model.device)
outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))
Evaluation
Evaluation settings: Temperature 0, shuffle seed 4242, full test datasets, HuggingFace Transformers inference (not vLLM), HRM builtin chat template. MATH limited to 500 samples due to judge grading bottleneck.
Key finding: 4-bit AWQ retains strong quality across all 17 benchmarks despite 34% size reduction. Average accuracy drop is just 1.4 points vs paper. On 3 benchmarks (BoolQ, MATH, PIQA-da) AWQ matches or exceeds the paper baseline.
English benchmark results (Best scores in bold.)
| Model | BoolQ (Acc) | Winogrande (Acc) | Hellaswag (Acc) | MMLU (Acc) | ARC-C (Acc) | DROP (F1) | GovRep. (R1) | Avg. |
|---|---|---|---|---|---|---|---|---|
| Mimir 1B (bf16) | 87.8 | 73.5 | 67.3 | 57.5 | 81.6 | 83.1 | 32.0 | 69.0 |
| Mimir 1B (AWQ FP4) | 87.9 | 71.8 | 66.4 | 57.6 | 80.5 | 78.3 | — | — |
| HRM-Text 1B | 87.5 | 70.4 | 60.4 | 58.7 | 82.2 | 78.1 | 25.4 | 66.1 |
| Qwen 3.5 0.8B | 69.8 | 48.9 | 37.0 | 51.5 | 68.4 | 45.2 | 32.5 | 50.5 |
| Gemma 3 1B | 62.4 | 49.1 | 30.6 | 37.5 | 43.5 | 7.0 | 29.5 | 37.1 |
| OLMo 2 1B | 67.2 | 51.0 | 42.4 | 41.6 | 48.1 | 12.4 | 37.7 | 42.9 |
| Qwen 3.5 2B | 80.8 | 53.4 | 64.6 | 62.8 | 82.7 | 31.3 | 31.5 | 58.2 |
| SmolLM3 3B | 84.3 | 60.3 | 65.1 | 60.2 | 79.5 | 54.0 | 38.1 | 63.1 |
| Qwen 3.5 4B | 87.0 | 70.0 | 83.2 | 75.8 | 92.9 | 48.0 | 27.9 | 69.3 |
| Gemma 4 E2B | 64.1 | 56.7 | 55.6 | 59.3 | 69.8 | 57.3 | 33.6 | 56.6 |
| Gemma 4 E2B (think) | 83.4 | 63.0 | 55.8 | 72.0 | 86.8 | 70.8 | 34.7 | 66.6 |
Math & Code benchmark results (Best scores in bold.)
| Model | GSM8K (Acc) | MATH (Acc) | HumanEval (Acc) | MBPP (Acc) | Avg. |
|---|---|---|---|---|---|
| Mimir 1B (bf16) | 89.9 | 45.8 | 56.7 | — | 64.1 |
| Mimir 1B (AWQ FP4) | 89.8 | 49.8 | 54.3 | 52.5 | — |
| HRM-Text 1B | 84.8 | 56.0 | 0.0 | — | 46.9 |
| Qwen 3.5 0.8B | 49.1 | 36.1 | 30.5 | — | 38.6 |
| Gemma 3 1B | 49.7 | 37.2 | 42.7 | — | 43.2 |
| OLMo 2 1B | 59.4 | 18.8 | 15.9 | — | 31.4 |
| Qwen 3.5 2B | 73.7 | 55.7 | 47.6 | — | 59.0 |
| SmolLM3 3B | 80.0 | 62.2 | 61.6 | — | 67.9 |
| Qwen 3.5 4B | 60.5 | 56.5 | 78.0 | — | 65.0 |
| Gemma 4 E2B | 88.3 | 64.2 | 73.8 | — | 75.4 |
| Gemma 4 E2B (think) | 90.3 | 49.1 | 72.0 | — | 70.5 |
Danish benchmark results (Best scores in bold.)
| Model | Angry Tweets (Acc) | DaLA (F1) | GEC (EM) | PIQA (Acc) | Daisy (EM) | WikiQA (EM) | WMT (chrF) | N.News (chrF) | IFEval (Acc) | Hellaswag-DA (Acc) | Gen. Tale. (Mgf) | Avg. |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mimir 1B (bf16) | 67.4 | 96.1 | 85.6 | 53.7 | 9.6 | 66.8 | 53.9 | 35.87 | 63.9 | 35.3 | — | 56.8 |
| Mimir 1B (AWQ FP4) | — | 95.8 | 92.9 | 53.7 | 7.6 | 66.4 | 53.6 | — | — | — | 30.0 | — |
| HRM-Text 1B | 42.4 | 26.7 | 0.5 | 13.0 | 0.0 | 34.9 | 25.4 | 26.76 | 18.5 | 28.8 | — | 21.7 |
| Qwen 3.5 0.8B | 53.8 | 51.0 | 0.7 | 56.5 | 0.7 | 41.6 | 37.8 | 35.30 | 39.6 | 25.0 | — | 34.2 |
| Gemma 3 1B | 54.4 | 41.0 | 3.3 | 72.2 | 1.4 | 42.6 | 45.1 | 35.56 | 47.2 | 24.8 | — | 36.8 |
| OLMo 2 1B | 33.6 | 48.7 | 0.2 | 75.0 | 0.0 | 8.4 | 30.0 | 33.77 | 32.5 | 26.7 | — | 28.9 |
| Qwen 3.5 2B | 61.6 | 36.4 | 8.0 | 25.0 | 2.5 | 49.4 | 45.6 | 34.85 | 56.1 | 24.7 | — | 34.4 |
| SmolLM3 3B | 63.2 | 33.5 | 3.3 | 51.9 | 2.2 | 0.3 | 37.3 | 35.98 | 49.8 | 40.1 | — | 31.7 |
| Qwen 3.5 4B | 69.1 | 50.1 | 42.6 | 70.4 | 4.7 | 57.1 | 52.1 | 37.03 | 73.7 | 34.7 | — | 49.2 |
| Gemma 4 E2B | 64.6 | 56.7 | 36.9 | 46.3 | 5.6 | 44.1 | 55.2 | 35.67 | 75.5 | 25.6 | — | 44.6 |
| Gemma 4 E2B (think) | 67.7 | 66.8 | 23.4 | 63.9 | 5.1 | 59.3 | 56.0 | 36.30 | 81.2 | 39.0 | — | 49.9 |
| Munin-Apertus 8B | 60.6 | 46.1 | 42.1 | 81.5 | 12.5 | 49.9 | 55.8 | 30.30 | 53.0 | 24.5 | — | 45.6 |
| Munin-Mistral 8B | 61.3 | 48.8 | 26.4 | 76.9 | 8.4 | 48.4 | 51.8 | 32.92 | 67.8 | 33.6 | — | 45.6 |
| Munin-Qwen 9B | 69.1 | 60.6 | 11.4 | 38.9 | 5.4 | 55.7 | 56.1 | 35.89 | 71.8 | 34.3 | — | 43.9 |
Model details
| Architecture | Parameters | Hidden size | Layers | Attention heads | Vocab size | Context length | Training steps | Tokens per epoch | License |
|---|---|---|---|---|---|---|---|---|---|
| HRM-Text | ~1B | 1,536 | 16 | 12 | 262,144 | 4,096 | 1,750,000 | ~70.5B | Apache 2.0 |
Technical Report
Training was performed using a fork of HRM-Text. Further details are provided in our technical report here.
Memorisation audit
We run two audits independently from each other across four data categories: synthetic instruction-tuning data derived from agreement-backed sources (A), instruction-tuning data from Hugging Face sources with uncertain opt-out status (B), instruction-tuning data from Hugging Face sources with high confidence of no applicable opt-out (C), and other low-risk synthetic and reasoning post-training datasets (D). All remaining datasets are either covered by open licenses, in the public domain, or synthetic datasets derived from such openly licensed or public domain datasets. Our memorisation audit thus covers all data sources where memorisation risk is meaningful. Full details are provided in the appendix of the technical report. Targeted prefix attacks yielded predominantly short matches, with only 0.00021%–0.00073% of training documents have been retrieved with 50 tokens or more across categories. Across all model-input evaluations in the second audit, only 0.000044% corresponded to coherent prose and 0.00000073% to expressive prose, with no high-priority copyright findings, indicating a generally low copyright-related memorisation risk under the tested scenarios.
Limitations
Mimir v1 was trained on Danish and English data only. It will likely have poor performance on other languages. The model has not been specifically aligned for safety and may reflect social biases present in its training data.
License
This model is released under the Apache License 2.0.
See the full license text in LICENSE.
Project partners & funding
The development of Mimir v1 was performed in close collaboration between University of Southern Denmark, Aarhus University, University of Copenhagen and the Alexandra Institute, as part of Danish Foundation Models.
Funding was provided by the Ministry of Science, Higher Education and Digital Affairs.
How to cite
@misc{mimir-v1,
title = {DFM Mimir v1: An Open HRM Delivering Frontier Performance at 1B Parameters Using Only Permissible Post-Training Data},
author = {Schneider-Kamp, Peter and Nielsen, Jacob and Barmina, Gicanluca and Enevoldsen, Kenneth and Poech, Lukas Galke},
year = {2026},
url = {https://huggingface.co/danish-foundation-models/HRM-Mimir-v1}
}
- Downloads last month
- 19
Model tree for schneiderkamplab/DFM-Mimir-AWQ-FP4
Base model
danish-foundation-models/DFM-Mimir