OLMo 2 7B Instruct

Allen Institute for AI · allen-ai/olmo-2-1124-7b-instruct

ProviderAllen Institute for AI
Familyolmo
Typellm-chat
Statusactive
Released2024-12-18
Updated2025-01-06
Parameters7.3B
Open weightsyes

Lineage

RelationshipModel
Is a finetune ofallenai/OLMo-2-1124-7B-DPO

What it fits on

Ordered by predicted speed. Bandwidth sets decode rate; memory decides whether it runs at all.

Every figure here is computed, not measured. Fit is weights at each quantisation against device memory, with a 25% allowance for the KV cache, activations and the OS. Decode rate is the memory-bandwidth roofline at 70% efficiency. Nobody has run this model on these devices.
DeviceMemoryBandwidthBest qualityWeightstok/sFastest
Cerebras WSE-344.0 GB21000000.0 GB/sbf1614.6 GB~1007040.1~4028160.2 q4
NVIDIA GB200 Grace Blackwell Superchip372.0 GB16000.0 GB/sbf1614.6 GB~767.3~3069.1 q4
AMD Instinct MI355X288.0 GB8000.0 GB/sbf1614.6 GB~383.6~1534.5 q4
NVIDIA B200180.0 GB7700.0 GB/sbf1614.6 GB~369.2~1477.0 q4
Google TPU7x (Ironwood)192.0 GB7380.0 GB/sbf1614.6 GB~353.9~1415.6 q4
AMD Instinct MI325X256.0 GB6000.0 GB/sbf1614.6 GB~287.7~1150.9 q4
AMD Instinct MI300X192.0 GB5300.0 GB/sbf1614.6 GB~254.2~1016.6 q4
NVIDIA H200 SXM141.0 GB4800.0 GB/sbf1614.6 GB~230.2~920.7 q4
NVIDIA H100 NVL94.0 GB3900.0 GB/sbf1614.6 GB~187.0~748.1 q4
NVIDIA H100 SXM80.0 GB3350.0 GB/sbf1614.6 GB~160.6~642.6 q4
AMD Instinct MI250X128.0 GB3200.0 GB/sbf1614.6 GB~153.5~613.8 q4
Google TPU v5p95.0 GB2765.0 GB/sbf1614.6 GB~132.6~530.4 q4
NVIDIA A100 80GB SXM80.0 GB2039.0 GB/sbf1614.6 GB~97.8~391.1 q4
NVIDIA H100 PCIe80.0 GB2000.0 GB/sbf1614.6 GB~95.9~383.6 q4
NVIDIA GeForce RTX 509032.0 GB1792.0 GB/sbf1614.6 GB~85.9~343.7 q4
Google TPU v6e (Trillium)32.0 GB1638.0 GB/sbf1614.6 GB~78.5~314.2 q4
AMD Instinct MI21064.0 GB1600.0 GB/sbf1614.6 GB~76.7~306.9 q4
NVIDIA A100 40GB SXM40.0 GB1555.0 GB/sbf1614.6 GB~74.6~298.3 q4
Google TPU v432.0 GB1200.0 GB/sbf1614.6 GB~57.5~230.2 q4
NVIDIA GeForce RTX 409024.0 GB1008.0 GB/sbf1614.6 GB~48.3~193.4 q4
AMD Radeon RX 7900 XTX24.0 GB960.0 GB/sbf1614.6 GB~46.0~184.1 q4
NVIDIA GeForce RTX 508016.0 GB960.0 GB/sfp87.3 GB~92.1~184.1 q4
NVIDIA GeForce RTX 309024.0 GB936.0 GB/sbf1614.6 GB~44.9~179.5 q4
NVIDIA L40S48.0 GB864.0 GB/sbf1614.6 GB~41.4~165.7 q4
Apple M3 Ultra512.0 GB819.0 GB/sbf1614.6 GB~39.3~157.1 q4
AMD Radeon RX 7900 XT20.0 GB800.0 GB/sbf1614.6 GB~38.4~153.5 q4
Apple M2 Ultra192.0 GB800.0 GB/sbf1614.6 GB~38.4~153.5 q4
Google TPU v5e16.0 GB800.0 GB/sfp87.3 GB~76.7~153.5 q4
NVIDIA GeForce RTX 4080 SUPER16.0 GB736.0 GB/sfp87.3 GB~70.6~141.2 q4
NVIDIA GeForce RTX 4070 Ti SUPER16.0 GB672.0 GB/sfp87.3 GB~64.5~128.9 q4
AMD Radeon RX 9070 XT16.0 GB640.0 GB/sfp87.3 GB~61.4~122.8 q4
Apple M4 Max128.0 GB546.0 GB/sbf1614.6 GB~26.2~104.7 q4
Apple M1 Max64.0 GB400.0 GB/sbf1614.6 GB~19.2~76.7 q4
Apple M2 Max96.0 GB400.0 GB/sbf1614.6 GB~19.2~76.7 q4
Apple M3 Max128.0 GB400.0 GB/sbf1614.6 GB~19.2~76.7 q4
NVIDIA GeForce RTX 3060 12GB12.0 GB360.0 GB/sfp87.3 GB~34.5~69.1 q4
NVIDIA L424.0 GB300.0 GB/sbf1614.6 GB~14.4~57.5 q4
NVIDIA GeForce RTX 4060 Ti 16GB16.0 GB288.0 GB/sfp87.3 GB~27.6~55.2 q4
Apple M4 Pro64.0 GB273.0 GB/sbf1614.6 GB~13.1~52.4 q4
Apple M432.0 GB120.0 GB/sbf1614.6 GB~5.8~23.0 q4

Where it runs

Taken from the card's availability section.

PlatformId on that platform
Hugging Faceallenai/OLMo-2-1124-7B-Instruct

Reported benchmark scores

This card reports no benchmark scores yet.

BenchmarkCatalogue standingScore

Data

This card as JSON · See it in the graph · Edit on GitHub