INDEX

Explanations

phrases related to deducing or relying on information

oai_token-act-pair · gpt-3.5-turbo

references to decisions or conclusions that are based on specific criteria or data

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

anamo

-0.78

pload

-0.75

scribe

-0.74

rolet

-0.71

erers

-0.71

anke

-0.70

apo

-0.70

asking

-0.70

robe

-0.70

joy

-0.70

POSITIVE LOGITS

 assumptions

1.08

 criteria

1.04

 assumption

1.04

 observations

1.03

 principles

0.99

 evaluations

0.96

 feedback

0.95

 whims

0.95

 premise

0.91

 intuition

0.90

Activations Density 0.386%

phrases related to deducing or relying on information

references to decisions or conclusions that are based on specific criteria or data

No Comments

No Known Activations