INDEX

Explanations

phrases related to breaks or recesses

oai_token-act-pair · gpt-3.5-turbo

mentions of congressional or parliamentary recesses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

odka

-0.72

gged

-0.70

 Solo

-0.68

Reply

-0.64

selage

-0.61

source

-0.60

gging

-0.60

FORE

-0.60

âĤ¬

-0.59

 traff

-0.58

POSITIVE LOGITS

 recess

1.17

es

1.01

ional

0.90

shire

0.88

ively

0.87

icut

0.87

 newcom

0.79

 appointments

0.79

entimes

0.79

wich

0.78

Activations Density 0.007%

phrases related to breaks or recesses

mentions of congressional or parliamentary recesses

No Comments

No Known Activations