---

library_name: transformers

language:

• en

pipeline_tag: text-generation

license: apache-2.0

base_model: JetBrains/Mellum2-12B-A2.5B-Instruct

tags:

• llama-cpp
• gguf-my-repo

model-index:

• name: Mellum2 Instruct

results:

- task:

type: text-generation

name: Text Generation

dataset:

name: LiveCodeBench v6

type: livecodebench

metrics:

- type: pass@1

value: 37.2

name: pass@1

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: EvalPlus (HumanEval+ / MBPP+ mean)

type: evalplus

metrics:

- type: pass@1

value: 78.4

name: pass@1

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: MultiPL-E (7 languages)

type: multipl-e

metrics:

- type: pass@1

value: 67.1

name: pass@1

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: BFCL v3

type: bfcl

metrics:

- type: acc

value: 66.3

name: accuracy

verified: false

- type: acc

value: 44.2

name: accuracy

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: AIME 2025+2026 (mean, 30 questions each)

type: aime

metrics:

- type: exact_match

value: 41.7

name: exact match

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: GSM-Plus

type: gsm-plus

metrics:

- type: exact_match

value: 80.5

name: exact match

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: MMLU-Redux

type: mmlu-redux

metrics:

- type: acc

value: 78.1

name: accuracy

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: GPQA Diamond

type: gpqa

metrics:

- type: acc

value: 40.9

name: accuracy

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: IFEval (prompt-level strict accuracy)

type: ifeval

metrics:

- type: acc

value: 75.8

name: accuracy

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: MixEval

type: mixeval

metrics:

- type: acc

value: 62.2

name: accuracy

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: BS-Bench (detection rate)

type: bs-bench

metrics:

- type: detection_rate

value: 18.0

name: detection rate

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: HarmBench (harmful rate, lower is better)

type: harmbench

metrics:

- type: harmful_rate

value: 23.1

name: harmful rate

verified: false

- task:

type: text-generation

name: Text Generation

dataset:

name: XSTest (safe compliance)

type: xstest

metrics:

- type: safe_compliance

value: 81.2

name: safe compliance

verified: false

---

TobDeBer/Mellum2-12B-A2.5B-Instruct-Q3_K_L-GGUF

This model was converted to GGUF format from JetBrains/Mellum2-12B-A2.5B-Instruct using llama.cpp via the ggml.ai's GGUF-my-repo space.

Refer to the original model card for more details on the model.

Use with llama.cpp

Install llama.cpp through brew (works on Mac and Linux)

brew install llama.cpp

Invoke the llama.cpp server or the CLI.

CLI:

llama-cli --hf-repo TobDeBer/Mellum2-12B-A2.5B-Instruct-Q3_K_L-GGUF --hf-file mellum2-12b-a2.5b-instruct-q3_k_l.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo TobDeBer/Mellum2-12B-A2.5B-Instruct-Q3_K_L-GGUF --hf-file mellum2-12b-a2.5b-instruct-q3_k_l.gguf -c 2048

Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.

git clone https://github.com/ggerganov/llama.cpp

Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).

cd llama.cpp && LLAMA_CURL=1 make

Step 3: Run inference through the main binary.

./llama-cli --hf-repo TobDeBer/Mellum2-12B-A2.5B-Instruct-Q3_K_L-GGUF --hf-file mellum2-12b-a2.5b-instruct-q3_k_l.gguf -p "The meaning to life and the universe is"

or

./llama-server --hf-repo TobDeBer/Mellum2-12B-A2.5B-Instruct-Q3_K_L-GGUF --hf-file mellum2-12b-a2.5b-instruct-q3_k_l.gguf -c 2048