Four-way 70B merge
According to the model card and hub tags, inputs include Hermes-3-Llama-3.1-70B, Dracarys-Llama-3.1-70B-Instruct, SauerkrautLM-70b-Instruct, and Meta-Llama-3.1-70B-Instruct.
Open Source Model Profile · gbueno86
Brinebreath-Llama-3.1-70B
Brinebreath-Llama-3.1-70B is a 70.55B-parameter Llama merge from gbueno86. Its model card describes merging Hermes 3, Dracarys, SauerkrautLM, and Meta-Llama-3.1 variants with publisher-run MMLU-PRO checks.
- Publisher
- gbueno86
- Task
- text-generation
- Model type
- llama
- License
- llama3.1
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
Brinebreath-Llama-3.1-70B is published by gbueno86 as a text-generation model. The captured configuration identifies LlamaForCausalLM with model type llama, and Safetensors metadata reports 70,553,706,496 parameters. According to the model card, it was assembled with MergeKit-style merges of four Llama-3.1-70B lines.
Recorded capabilities
Publisher MMLU-PRO table
The publisher reports 49.0% for its Q4_0 file versus 42.0% for Meta-Llama-3.1-70B-Instruct Q4_0 on a 100-question MMLU-PRO check.
Manual test notes
According to the model card, manual checks cover common-sense, programming, math, and story cases with per-case OK and KO marks.
Use cases in the source record
- Conversational text generation with a large merged Llama-3.1-70B setup.
- Merge-comparison work that references the publisher's MMLU-PRO and manual test tables.
Limitations and unknowns
- MMLU-PRO figures are small-sample publisher tests, 100 questions overall and 20 per category, not Ethen-measured results.
- No context-window value was extracted from this record.
- No independently verified evaluation results were extracted from this record.
- Provider state is historical snapshot data, not independently refreshed current availability.
Source and provenance
Source: gbueno86/Brinebreath-Llama-3.1-70B
Captured: Unknown. Processed: 2026-09-07T19:34:45.180205+00:00.
Brinebreath-Llama-3.1-70B I made this since I started having some problems with Cathallama. This seems to behave well during some days testing. Notable Performance 7% overall success rate increase on MMLU-PRO over LLaMA 3.1 70b at Q4_0 Strong performance in MMLU-PRO categories overall Great performance during manual testing Creation workflow Models merged meta-llama/Meta-Llama-3.1-70B-Instruct NousResearch/Hermes-3-Llama-3.1-70B abacusai/Dracarys-Llama-3.1-70B-Instruct VAGOsolutions/Llama-3.1-SauerkrautLM-70b-Instruct flowchart TD A[Hermes 3] -->|Merge with| B[Meta-Llama-3.1] C[Dracarys] -->|Merge with| D[Meta-Llama-3.1] B -->| | E[…
F001F002F003F004F005F006F007F008F009F010F011F012F013F014