INDEX

Explanations

instances of events or actions happening at a specific point in time

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "when" to indicate time references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ouble

-0.84

ictive

-0.80

whatever

-0.74

iosyncr

-0.73

gem

-0.72

uce

-0.69

omal

-0.68

arling

-0.66

kaya

-0.65

educ

-0.64

POSITIVE LOGITS

soever

1.12

he

1.04

 they

1.03

 asked

1.02

 confronted

0.99

 contacted

0.93

she

0.91

 faced

0.85

 interviewed

0.82

 approached

0.81

Activations Density 0.103%

instances of events or actions happening at a specific point in time

occurrences of the word "when" to indicate time references

No Comments

No Known Activations