metadata
license: apache-2.0
language:
- en
library_name: transformers
datasets:
- allenai/dolma3_mix-5.5T-1125
tags:
- llama-cpp
- gguf-my-repo
base_model: allenai/Olmo-3-1125-32B
model-index:
- name: Olmo-3-1125-32B
results:
- task:
type: text-generation
dataset:
name: Benchmarks
type: benchmark
metrics:
- type: olmo_3_eval_math
value: 61.6
name: Olmo 3-Eval Math
- type: bigcodebench
value: 43.9
name: BigCodeBench
- type: humaneval
value: 66.5
name: HumanEval
- type: deepseek_leetcode
value: 1.9
name: DeepSeek LeetCode
- type: ds_1000
value: 29.7
name: DS 1000
- type: mbpp
value: 60.2
name: MBPP
- type: multipl_humaneval
value: 35.9
name: MultiPL HumanEval
- type: multipl_mbppp
value: 41.8
name: MultiPL MBPPP
- type: olmo_3_eval_code
value: 40
name: Olmo 3-Eval Code
- type: arc_mc
value: 94.7
name: ARC MC
- type: mmlu_stem
value: 70.8
name: MMLU STEM
- type: medmcqa_mc
value: 57.6
name: MedMCQA MC
- type: medqa_mc
value: 53.8
name: MedQA MC
- type: sciq_mc
value: 95.5
name: SciQ MC
- type: olmo_3_eval_mc_stem
value: 74.5
name: Olmo 3-Eval MC_STEM
- type: mmlu_humanities
value: 78.3
name: MMLU Humanities
- type: mmlu_social_sci.
value: 83.9
name: MMLU Social Sci.
- type: mmlu_other
value: 75.1
name: MMLU Other
- type: csqa_mc
value: 82.3
name: CSQA MC
- type: piqa_mc
value: 85.6
name: PIQA MC
- type: socialiqa_mc
value: 83.9
name: SocialIQA MC
- type: coqa_gen2mc_mc
value: 96.4
name: CoQA Gen2MC MC
- type: drop_gen2mc_mc
value: 87.2
name: DROP Gen2MC MC
- type: jeopardy_gen2mc_mc
value: 92.3
name: Jeopardy Gen2MC MC
- type: naturalqs_gen2mc_mc
value: 78
name: NaturalQs Gen2MC MC
- type: squad_gen2mc_mc
value: 98.2
name: SQuAD Gen2MC MC
- type: olmo_3_eval_mc_non_stem
value: 85.6
name: Olmo 3-Eval MC_Non-STEM
- type: hellaswag_rc
value: 84.8
name: HellaSwag RC
- type: winogrande_rc
value: 90.3
name: Winogrande RC
- type: lambada
value: 75.7
name: Lambada
- type: basic_skills
value: 93.5
name: Basic Skills
- type: drop
value: 81
name: DROP
- type: jeopardy
value: 75.3
name: Jeopardy
- type: naturalqs
value: 48.7
name: NaturalQs
- type: squad
value: 94.5
name: SQuAD
- type: coqa
value: 74.1
name: CoQA
- type: olmo_3_eval_genqa
value: 79.8
name: Olmo 3-Eval GenQA
- type: bbh
value: 77.6
name: BBH
- type: mmlu_pro_mc
value: 49.6
name: MMLU Pro MC
- type: deepmind_math
value: 30.1
name: Deepmind Math
- type: lbpp
value: 21.7
name: LBPP
source:
url: https://huggingface.co/allenai/Olmo-3-1125-32B
name: Model README
IonizedLeaf/Olmo-3-1125-32B-Q8_0-GGUF
This model was converted to GGUF format from allenai/Olmo-3-1125-32B 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 IonizedLeaf/Olmo-3-1125-32B-Q8_0-GGUF --hf-file olmo-3-1125-32b-q8_0.gguf -p "The meaning to life and the universe is"
Server:
llama-server --hf-repo IonizedLeaf/Olmo-3-1125-32B-Q8_0-GGUF --hf-file olmo-3-1125-32b-q8_0.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 IonizedLeaf/Olmo-3-1125-32B-Q8_0-GGUF --hf-file olmo-3-1125-32b-q8_0.gguf -p "The meaning to life and the universe is"
or
./llama-server --hf-repo IonizedLeaf/Olmo-3-1125-32B-Q8_0-GGUF --hf-file olmo-3-1125-32b-q8_0.gguf -c 2048