INDEX

Explanations

terms related to specific geographic locations or areas

oai_token-act-pair · gpt-3.5-turbo

references to areas or regions, particularly in a mathematical or spatial context

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

inn

-0.87

vous

-0.87

lio

-0.77

aldi

-0.70

bones

-0.70

enstein

-0.69

essed

-0.68

sing

-0.67

ging

-0.67

estamp

-0.67

POSITIVE LOGITS

 bounded

0.81

 area

0.79

 malls

0.76

 surrounding

0.76

rug

0.75

 areas

0.75

 frequ

0.74

 adjacent

0.74

 encomp

0.72

 mall

0.71

Activations Density 0.050%

terms related to specific geographic locations or areas

references to areas or regions, particularly in a mathematical or spatial context

No Comments

No Known Activations