INDEX

Explanations

phrases related to verbal communication about policies, opinions, or events

oai_token-act-pair · gpt-3.5-turbo

discussions and mentions of verbal communication

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

pps

-0.77

 ILCS

-0.71

ertain

-0.63

ective

-0.63

cial

-0.60

 Peaks

-0.59

wick

-0.58

eele

-0.58

inet

-0.58

hook

-0.58

POSITIVE LOGITS

 extensively

1.25

 passionately

1.07

 about

1.04

 nostalg

1.01

 eloqu

1.00

 favorably

0.99

 openly

0.98

 endlessly

0.98

 dispar

0.96

 omin

0.96

Activations Density 0.092%

phrases related to verbal communication about policies, opinions, or events

discussions and mentions of verbal communication

No Comments

No Known Activations