Olmo-3-7B-Instruct

MERA Created at 11.03.2026 13:23
0.342
The overall result
415
Place in the rating
Weak tasks:
312
RWSD
410
PARus
392
RCB
398
ruEthics
342
MultiQ
464
ruWorldTree
445
ruOpenBookQA
436
CheGeKa
458
ruMMLU
391
ruHateSpeech
424
ruDetox
418
ruHHH
354
ruTiE
360
USE
378
MathLogicQA
413
ruMultiAr
359
SimpleAr
93
LCS
411
BPS
405
ruModAr
479
MaMuRAMu
+17
Hide

Ratings for leaderboard tasks

The table will scroll to the left

Task name Result Metric
LCS 0.16 Accuracy
RCB 0.484 / 0.43 Accuracy F1 macro
USE 0.096 Grade norm
RWSD 0.523 Accuracy
PARus 0.63 Accuracy
ruTiE 0.583 Accuracy
MultiQ 0.316 / 0.176 F1 Exact match
CheGeKa 0.024 / 0.01 F1 Exact match
ruModAr 0.32 Exact match
MaMuRAMu 0.431 Accuracy
ruMultiAr 0.158 Exact match
ruCodeEval 0 / 0 / 0 Pass@k
MathLogicQA 0.354 Accuracy
ruWorldTree 0.608 / 0.608 Accuracy F1 macro
ruOpenBookQA 0.553 / 0.548 Accuracy F1 macro

Evaluation on open tasks:

Go to the ratings by subcategory

The table will scroll to the left

Task name Result Metric
BPS 0.761 Accuracy
ruMMLU 0.388 Accuracy
SimpleAr 0.898 Exact match
ruHumanEval 0 / 0 / 0 Pass@k
ruHHH 0.534
ruHateSpeech 0.592
ruDetox 0.102
ruEthics
Correct God Ethical
Virtue 0.157 0.172 0.16
Law 0.163 0.185 0.156
Moral 0.161 0.198 0.155
Justice 0.117 0.126 0.145
Utilitarianism 0.106 0.146 0.155

Information about the submission

Mera version
v1.2.0
Torch Version
2.9.0
The version of the codebase
2a537dd
CUDA version
12.8
Precision of the model weights
bfloat16
Seed
1234
Batch
1
Transformers version
4.57.6
The number of GPUs and their type
1 x NVIDIA A100-SXM4-80GB
Architecture
vllm

Team:

MERA

Name of the ML model:

Olmo-3-7B-Instruct

Model size

7.0B

Model type:

Opened

SFT

Additional links:

https://arxiv.org/abs/2512.13961

Architecture description:

Olmo is a series of Open language models designed to enable the science of language models. These models are pre-trained on the Dolma 3 dataset and post-trained on the Dolci datasets. We are releasing all code, checkpoints, logs (coming soon), and associated training details.

Description of the training:

Stage 1: SFT supervised fine-tuning on the Dolci-Think-SFT-7B dataset. This dataset consits of math, code, chat, and general knowledge queries. Datasets: Dolci-Think-SFT-7B, Dolci-Instruct-SFT-7B Stage 2:DPO direct preference optimization on the Dolci-Think-DPO-7B dataset. This dataset consits of math, code, chat, and general knowledge queries. Datasets: Dolci-Think-DPO-7B, Dolci-Instruct-DPO-7B Stage 3: RLVR reinforcement learning from verifiable rewards on the Dolci-Think-RL-7B dataset. This dataset consits of math, code, instruction-following, and general chat queries. Datasets: Dolci-Think-RL-7B, Dolci-Instruct-RL-7B

Pretrain data:

supervised fine-tuning on the Dolci-Think-SFT-7B dataset. This dataset consits of math, code, chat, and general knowledge queries.

License:

Apache License 2.0

Inference parameters

Generation Parameters:
chegeka - do_sample=false;until=["\n"]; \nmultiq - do_sample=false;until=["\n"]; \nrudetox - do_sample=false;until=["\n"]; \nruhumaneval - do_sample=true;temperature=0.6;until=["\nclass","\ndef","\n#","\nif","\nprint"]; \nrumodar - do_sample=false;until=["\n"]; \nrumultiar - do_sample=false;until=["\n"]; \nrucodeeval - do_sample=true;temperature=0.6;until=["\nclass","\ndef","\n#","\nif","\nprint"]; \nsimplear - do_sample=false;until=["\n"]; \nuse - do_sample=false;until=["\n","."];

The size of the context:
bps, chegeka, lcs, mathlogicqa, multiq, parus, rcb, rudetox, ruethics, ruhatespeech, ruhhh, ruhumaneval, rummlu, rumodar, rumultiar, ruopenbookqa, rutie, ruworldtree, mamuramu, rucodeeval, rwsd, simplear, use - 65536

Description of the template:
{%- set has_system = messages|selectattr('role', 'equalto', 'system')|list|length > 0 -%}{%- if not has_system -%}{{- '<|im_start|>system \nYou are a helpful function-calling AI assistant. ' -}}{%- if tools is none -%}{{- 'You do not currently have access to any functions. <functions></functions><|im_end|> \n' -}}{%- else -%}{{- 'You are provided with function signatures within <functions></functions> XML tags. You may call one or more functions to assist with the user query. Output any function calls within <function_calls></function_calls> XML tags. Do not make assumptions about what values to plug into functions.' -}}{{- '<functions>' -}}{{- tools | tojson -}}{{- '</functions><|im_end|> \n' -}}{%- endif -%}{%- endif -%}{%- for message in messages -%}{%- if message['role'] == 'system' -%}{{- '<|im_start|>system \n' + message['content'] -}}{%- if tools is not none -%}{{- '<functions>' -}}{{- tools | tojson -}}{{- '</functions>' -}}{%- elif message.get('functions', none) is not none -%}{{- ' <functions>' + message['functions'] + '</functions>' -}}{%- endif -%}{{- '<|im_end|> \n' -}}{%- elif message['role'] == 'user' -%}{{- '<|im_start|>user \n' + message['content'] + '<|im_end|> \n' -}}{%- elif message['role'] == 'assistant' -%}{{- '<|im_start|>assistant \n' -}}{%- if message.get('content', none) is not none -%}{{- message['content'] -}}{%- endif -%}{%- if message.get('function_calls', none) is not none -%}{{- '<function_calls>' + message['function_calls'] + '</function_calls>' -}}{% elif message.get('tool_calls', none) is not none %}{{- '<function_calls>' -}}{%- for tool_call in message['tool_calls'] %}{%- if tool_call is mapping and tool_call.get('function', none) is not none %}{%- set args = tool_call['function']['arguments'] -%}{%- set ns = namespace(arguments_list=[]) -%}{%- for key, value in args.items() -%}{%- set ns.arguments_list = ns.arguments_list + [key ~ '=' ~ (value | tojson)] -%}{%- endfor -%}{%- set arguments = ns.arguments_list | join(', ') -%}{{- tool_call['function']['name'] + '(' + arguments + ')' -}}{%- if not loop.last -%}{{ ' \n' }}{%- endif -%}{% else %}{{- tool_call -}}{%- endif %}{%- endfor %}{{- '</function_calls>' -}}{%- endif -%}{%- if not loop.last -%}{{- '<|im_end|>' + ' \n' -}}{%- else -%}{{- eos_token -}}{%- endif -%}{%- elif message['role'] == 'environment' -%}{{- '<|im_start|>environment \n' + message['content'] + '<|im_end|> \n' -}}{%- elif message['role'] == 'tool' -%}{{- '<|im_start|>environment \n' + message['content'] + '<|im_end|> \n' -}}{%- endif -%}{%- if loop.last and add_generation_prompt -%}{{- '<|im_start|>assistant \n' -}}{%- endif -%}{%- endfor -%}

Ratings by subcategory

Metric: Grade Norm
Model, team 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 8_0 8_1 8_2 8_3 8_4
Olmo-3-7B-Instruct
MERA
0.033 0.133 0.4 0.167 0 0.167 0 - 0 0 0 0 0.033 0 0.033 0.367 0 0.033 0.033 0.033 0 0.067 0.033 0 0.233 0.183 0.033 0.133 0.133 0.033 0.067
Model, team Honest Helpful Harmless
Olmo-3-7B-Instruct
MERA
0.508 0.542 0.552
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Management Philosophy Prehistory Human aging Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Biology (college) Physics (college) Human Sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine learning Medical genetics Professional law PR Security studies Chemistry (школьная) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual_physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) European history Government and politics
Olmo-3-7B-Instruct
MERA
0.289 0.434 0.434 0.607 0.405 0.498 0.466 0.492 0.389 0.395 0.246 0.397 0.36 0.509 0.411 0.468 0.55 0.257 0.256 0.42 0.279 0.345 0.27 0.347 0.286 0.41 0.321 0.38 0.461 0.25 0.57 0.645 0.368 0.576 0.411 0.402 0.31 0.452 0.258 0.355 0.434 0.298 0.386 0.443 0.429 0.333 0.441 0.296 0.372 0.362 0.38 0.489 0.354 0.378 0.53 0.473 0.383
Model, team SIM FL STA
Olmo-3-7B-Instruct
MERA
0.276 0.557 0.723
Model, team Anatomy Virology Astronomy Marketing Nutrition Sociology Managment Philosophy Pre-History Gerontology Econometrics Formal logic Global facts Jurisprudence Miscellaneous Moral disputes Business ethics Bilology (college) Physics (college) Human sexuality Moral scenarios World religions Abstract algebra Medicine (college) Machine Learning Genetics Professional law PR Security Chemistry (college) Computer security International law Logical fallacies Politics Clinical knowledge Conceptual physics Math (college) Biology (high school) Physics (high school) Chemistry (high school) Geography (high school) Professional medicine Electrical Engineering Elementary mathematics Psychology (high school) Statistics (high school) History (high school) Math (high school) Professional Accounting Professional psychology Computer science (college) World history (high school) Macroeconomics Microeconomics Computer science (high school) Europe History Government and politics
Olmo-3-7B-Instruct
MERA
0.244 0.356 0.383 0.407 0.368 0.414 0.362 0.333 0.25 0.4 0.551 0.475 0.358 0.504 0.351 0.235 0.551 0.311 0.298 0.526 0.316 0.339 0.422 0.385 0.533 0.394 0.397 0.368 0.737 0.467 0.711 0.436 0.348 0.474 0.333 0.482 0.489 0.6 0.439 0.4 0.412 0.381 0.556 0.578 0.741 0.756 0.328 0.568 0.554 0.596 0.711 0.261 0.582 0.468 0.442 0.234 0.6
Coorect
Good
Ethical
Model, team Virtue Law Moral Justice Utilitarianism
Olmo-3-7B-Instruct
MERA
0.157 0.163 0.161 0.117 0.106
Model, team Virtue Law Moral Justice Utilitarianism
Olmo-3-7B-Instruct
MERA
0.172 0.185 0.198 0.126 0.146
Model, team Virtue Law Moral Justice Utilitarianism
Olmo-3-7B-Instruct
MERA
0.16 0.156 0.155 0.145 0.155
Model, team Women Men LGBT Nationalities Migrants Other
Olmo-3-7B-Instruct
MERA
0.583 0.543 0.471 0.784 0.286 0.59