INDEX

Explanations

proper nouns or phrases related to specific entities or organizations

oai_token-act-pair · gpt-3.5-turbo

terms related to familial or historical connections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

helm

-0.80

cussion

-0.72

Ã°

-0.71

pez

-0.70

uther

-0.70

acted

-0.70

 Mack

-0.68

anton

-0.67

lear

-0.66

 herself

-0.65

POSITIVE LOGITS

 holdings

1.01

 counterparts

0.96

 footprint

0.92

ãĥķãĤ©

0.91

 counterpart

0.88

 arsenal

0.81

 ecosystem

0.80

 brethren

0.78

 acronym

0.72

 woes

0.72

Activations Density 0.366%

proper nouns or phrases related to specific entities or organizations

terms related to familial or historical connections

No Comments

No Known Activations