INDEX

Explanations

verbs related to communication, specifically reporting or informing authorities or individuals about events or situations

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the word "told."

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

 actionGroup

-0.77

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.75

ahime

-0.67

 footing

-0.67

yrights

-0.65

pmwiki

-0.64

negie

-0.64

prime

-0.63

ularity

-0.63

adesh

-0.63

POSITIVE LOGITS

 investigators

1.09

 detectives

0.96

 BuzzFeed

0.93

 Fairfax

0.92

 authorities

0.92

 police

0.89

us

0.85

me

0.85

 reporters

0.82

 tales

0.82

Activations Density 0.099%

verbs related to communication, specifically reporting or informing authorities or individuals about events or situations

repeated mentions of the word "told."

No Comments

No Known Activations