logo A fine-tune of unsloth/Llama-3.2-1B-Instruct on the m-a-p/SuperGPQA dataset.

Usage example

Set temperature as 0.0 for best results.

System prompt

You are a classifier. Categorize the following problem into discipline, field, and subfield in JSON format.

User prompt

Cotton and linen both readily catch fire. A batch of towels is composed of both cotton and linen, and is known to have caught fire. If it is known that the towels were ignited by a lit cigarette, which of the following arguments utilizes the most appropriate form of reasoning?

Assistant response

{"discipline": "Philosophy", "field": "Philosophy", "subfield": "Logic"}

Possible output options

Discipline

['Economics', 'Medicine', 'Law', 'Management', 'Sociology', 'Science', 'Philosophy', 'Military Science', 'History', 'Literature and Arts', 'Engineering', 'Agronomy', 'Education']

Field

['Crop Science', 'Mathematics', 'Nuclear Science and Technology', 'Chemical Engineering and Technology', 'Optical Engineering', 'Journalism and Communication', 'Food Science and Engineering', 'Information and Communication Engineering', 'Traditional Chinese Medicine', 'Geology', 'Aquaculture', 'Animal Husbandry', 'Electronic Science and Technology', 'Geophysics', 'Metallurgical Engineering', 'Architecture', 'Forestry Engineering', 'Oceanography', 'Materials Science and Engineering', 'Transportation Engineering', 'Electrical Engineering', 'Weapon Science and Technology', 'History', 'Geological Resources and Geological Engineering', 'Instrument Science and Technology', 'Naval Architecture and Ocean Engineering', 'Agricultural Engineering', 'Business Administration', 'Surveying and Mapping Science and Technology', 'Civil Engineering', 'Clinical Medicine', 'Art Studies', 'Forestry', 'Physics', 'Management Science and Engineering', 'Textile Science and Engineering', 'Environmental Science and Engineering', 'Philosophy', 'Musicology', 'Political Science', 'Aeronautical and Astronautical Science and Technology', 'Theoretical Economics', 'Psychology', 'Sociology', 'Power Engineering and Engineering Thermophysics', 'Applied Economics', 'Control Science and Engineering', 'Hydraulic Engineering', 'Biology', 'Law', 'Stomatology', 'Petroleum and Natural Gas Engineering', 'Mechanics', 'Astronomy', 'Systems Science', 'Chemistry', 'Mechanical Engineering', 'Computer Science and Technology', 'Pharmacy', 'Atmospheric Science', 'Physical Education', 'Mining Engineering', 'Military Science', 'Language and Literature', 'Public Health and Preventive Medicine', 'Public Administration', 'Physical Oceanography', 'Basic Medicine', 'Veterinary Medicine', 'Geography', 'Library, Information and Archival Management', 'Education']

Subfield

['Principles of Metallurgy', 'Thermal Energy Engineering', 'Relativity', 'Military Command and Information Systems', 'Clinical Laboratory Diagnostics', 'Literary History', 'Archaeology and Museology', 'Oncology', 'Computer Software and Theory', 'Physiology', 'Electrodynamics', 'Western Economics', 'Public Finance', 'Computer Architecture', 'Library and Archival Science', 'Agricultural Mechanization Engineering', 'Criminal Law', 'Theory of Curriculum and Instruction', 'Solid State Physics', 'Religious Studies', 'Electrochemistry', 'Finance', 'Food Biochemistry', 'Materials Processing Engineering', 'Antenna and Radio Communication', 'Geological Resources and Geological Engineering', 'Thermodynamics and Statistical Physics', 'Marine Biology', 'Non-ferrous Metallurgy', 'Animal Nutrition and Feed Science', 'Forest Engineering', 'Mechatronic Engineering', 'Marine Engineering', 'Chemical Transport Engineering', 'Philology and Bibliography', 'Solid Mechanics', 'Physical Chemistry', 'Medicinal Chemistry', 'Landscape Plants and Ornamental Horticulture', 'Vehicle Operation Engineering', 'Biophysics', 'Atomic and Molecular Physics', 'Political Science', 'Health Toxicology and Environmental Health', 'Labor Economics', 'Basic Stomatology', 'Cryptography', 'Harmony', 'Ecology', 'Polynomials and Series Expansions', 'Ordinary Differential Equations', 'Modern and Contemporary Chinese Literature', 'Human Geography', 'Fluid Physics', 'Social and Folklore Studies', 'Dance Studies', 'Pitch and Scales', 'Special Education', 'Mass Transport and Separation Process in Chemical Engineering', 'Digital Surveying and Remote Sensing Applications', 'Pharmaceutics', 'Literary Theory', 'Communication and Broadcasting', 'Anesthesiology', 'Military Law', 'Immunology', 'Pathology and Pathophysiology', 'Quantum Mechanics', 'Educational Technology and Principles', 'Structural Engineering', 'Pediatrics', 'Legal Theory and Legal History', 'Ship Mechanics and Design Principles', 'Cell Biology', 'Nuclear Energy and Reactor Technology', 'Heat Transfer', 'Contract Law', 'Inorganic Chemistry', 'Laser Technology', 'Textile Chemistry and Dyeing Engineering', 'Microbiology and Biochemical Pharmacy', 'Refrigeration and Cryogenic Engineering', 'Journalism and News Practice', 'Weapon Systems Science and Engineering', 'Urban Planning and Design', 'Physical Geography', 'Constitutional and Administrative Law', 'Theoretical Mechanics', 'Microelectronics and Solid-State Electronics', 'Physical Chemistry of Metallurgical Process', 'Information Management Science', 'Microbiology', 'Guidance, Navigation and Control', 'Quantitative Economics', 'Genetics', 'Traffic Information Engineering and Control', 'History and Theory of Journalism and Media Management', 'Polymer Physics', 'Management Science and Engineering', 'Astronomical Observation and Technology', 'Combinatorial Mathematics', 'Mathematical Analysis', 'Education Economics, Management and Social Security', 'Law and Social Governance', 'Environmental and Resource Protection', 'Historical Geography', 'Psychology', 'Instrumentation and Performance', 'Political Economy', 'Databases', 'Operations Research and Cybernetics', 'Music History, Education, and Technology', 'Fuzzy Mathematics', 'Nursing and Rehabilitation Medicine', 'Architectural History', 'Systems Science', 'Internal Medicine', 'Economic Statistics', 'Military Chemistry and Pyrotechnics', 'Psychiatry and Mental Health', 'Numerical Analysis', 'Astrophysics', 'Dynamic Meteorology', 'Mineralogy, Petrology, and Economic Geology', 'Physical Oceanography', 'Materials Physics and Chemistry', 'Manufacturing Automation', 'Drama and Opera Studies', 'Demography and Anthropology', 'Thermodynamics', 'Veterinary Medicine', 'Russian Language and Literature', 'Signal and Information Processing', 'Water conservancy and Hydropower Engineering', 'Group Theory', 'Animal Rearing and Breeding', 'Electromagnetic Field and Microwave Technology', 'Cartography and Geographic Information Engineering', 'Environmental Engineering', 'Design Arts', 'Mineral Processing Engineering', 'Space physics', 'Mining and Safety Engineering', 'Structural Geology', 'Poromechanics and Reservoir Physics', 'Physical Education and Training', 'Preschool Education', 'Epidemiology and Health Statistics', 'Pattern Recognition', 'Neurology', 'Stochastic Processes', 'Geodesy and Surveying Engineering', 'Textile Materials Science', 'Communication and Information Systems', 'Principles of Computer Organization', 'Solar System Science', 'Special Number Theory', 'Power Machinery and Engineering', 'Fluid Flow and Heat Transfer in Chemical Engineering', 'Advanced Algebra', 'Environmental Science', 'Geochemistry', 'Pathogen Biology', 'Internal Combustion Engineering', 'Oil and Gas Field Development and Storage & Transportation Engineering', 'Bridge and Tunnel Engineering', 'Logic', 'Transportation Planning and Management', 'Marine Chemistry', 'Geometry and Topology', 'Fluid Machinery and Engineering', 'Information Management and Communication', 'Discrete Mathematics', 'Otorhinolaryngology', 'Clinical Stomatology', 'Hydrogeology', 'Business and Accounting Management', 'Theoretical Optics', 'Radiation Protection and Nuclear Technology Applications', 'Dermatology and Venereology', 'Military Logistics and Equipment', 'Pharmaceutical Analysis', 'Graph Theory', 'Engineering Thermophysics', 'Traditional Chinese Medicine Theory', 'Aeronautical and Astronautical Science and Technology', 'Analytical Chemistry', 'Hydraulics and Hydrology', 'Formal Languages', 'Surgery', 'Optoelectronic Technology', 'Agricultural Environment and Soil-Water Engineering', 'Statistical Mechanics', 'Road and Railway Engineering', 'Forest Cultivation and Genetic Breeding', 'Organic Chemistry', 'Solid Earth Geophysics', 'Composition', 'Operating Systems', 'Subatomic and Atomic Physics', 'Wood Science and Technology', 'Sports Humanities and Sociology', 'Power Systems and Automation', 'Biochemistry and Molecular Biology', 'Pharmacology', 'Fundamental Mathematics', 'Data Structures', 'Geotechnical Engineering', 'Advanced Programming Languages', 'Control Theory and Control Engineering', 'Broadcasting and Television Art', 'Semiconductor Physics', 'Engineering Fluid Mechanics', 'Polymer Chemistry and Physics', 'Urban Infrastructure Engineering', 'Electrical Theory and New Technologies', 'Functions of Real Variables', 'Crop Science', 'Maternal, Child and Adolescent Health', 'Fine Arts', 'Forensic Medicine', 'Economic History', 'Land Resource Management and Administrative Management', 'Obstetrics and Gynecology', 'Food Processing and Storage Engineering', 'Philosophy of Science and Technology', 'Stellar and Interstellar Evolution', 'Cosmology', 'Probability and Statistics', 'Architectural Design and Theory', 'Botany', 'Aquaculture', 'Principles of Seismic Exploration', 'Paleontology and Stratigraphy', 'Power Electronics and Electrical Drives', 'Human Anatomy and Histology-Embryology', 'Classical Chinese Literature', 'International Law', 'Ophthalmology', 'High Voltage and Insulation Technology', 'World History', 'International Trade', 'Military Thought and History', 'Industrial Economics', 'Linguistics and Applied Linguistics', 'Geriatric Medicine', 'Computational Mathematics', 'Functions of Complex Variables', 'Fundamentals of Dynamics and Control', 'Tourism Management and Technological Economics Management', 'Circuits and Systems', 'Instrument Science and Technology', 'Underwater Acoustics', 'Social Medicine and Health Management', 'Atmospheric Physics and Atmospheric Environment', 'Ethics', 'Imaging and Nuclear Medicine', 'Optical Fiber Communication', 'Film Studies', 'Military Management', 'French Language and Literature', 'Civil and Commercial Law', 'Meteorology', 'Applied Optics', 'Particle and Nuclear Physics', 'Acoustics', 'Zoology', 'Nutrition and Food Hygiene', 'Computer Networks', 'Theoretical Fluid Mechanics', 'Traditional Chinese Pharmacy', 'Sports Science and Medicine', 'Number Theory', 'Elements of Chemical Reaction Engineering', 'Rigid Body Mechanics', 'Emergency Medicine', 'Musical Forms and Analysis', 'Communication Principles', 'Philosophical Aesthetics', 'Radiation Medicine', 'Radiochemistry', 'National and Defense Economics', 'Procedural Law', 'Iron and Steel Metallurgy', 'Traditional Chinese Health Preservation']

Model Details

  • Base Model: unsloth/Llama-3.2-1B-Instruct
  • Parameter Count: 1,235,814,400
  • Precision: torch.bfloat16

Hardware

  • GPU: NVIDIA RTX PRO 6000 Blackwell Server Edition
  • Announced: Mar 17th, 2025
  • Release Date: Mar 18th, 2025
  • Memory Type: GDDR7
  • Bandwidth: 1.79 TB/s
  • Memory Size: 96 GB
  • Memory Bus: 512 bit
  • Shading Units: 24064
  • TDP: 600W

Training Settings

PEFT

  • Rank: 32
  • LoRA alpha: 64
  • Modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
  • Gradient checkpointing: unsloth

SFT

  • Epoch: 2
  • Batch size: 32
  • Gradient Accumulation steps: 1
  • Warmup ratio: 0.05
  • Learning rate: 0.0004
  • Optimizer: adamw_torch_fused
  • Learning rate scheduler: cosine

Training stats

  • Date: 2026-03-23T12:12:51.833430
  • Peak VRAM usage: 17.775 GB
  • Global step: 1576
  • Training runtime (seconds): 623.1828
  • Average training loss: 0.07477703140244871
  • Final validation loss: 0.05316569283604622

Framework versions

  • Unsloth: 2026.3.10
  • TRL: 0.22.2
  • Transformers: 4.56.2
  • Pytorch: 2.10.0+cu128
  • Datasets: 4.8.3
  • Tokenizers: 0.22.2

License

This model is released under the Llama3 license. See the Terms of Use for details.

Downloads last month
477
Safetensors
Model size
1B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for kth8/Llama-3.2-1B-Instruct-SuperGPQA-Classifier

Finetuned
(420)
this model
Quantizations
1 model

Dataset used to train kth8/Llama-3.2-1B-Instruct-SuperGPQA-Classifier