INDEX

Explanations

phrases related to reasons or purposes for actions

oai_token-act-pair · gpt-3.5-turbo

phrases related to consequences or purposes of actions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

awed

-0.71

 Eleven

-0.68

©¶æ¥µ

-0.66

falls

-0.62

joined

-0.61

 Hopkins

-0.61

riad

-0.60

icol

-0.60

 THREE

-0.60

antine

-0.60

POSITIVE LOGITS

 amusement

1.05

 assisting

1.04

 entertainment

1.03

 aiding

0.94

 facilitating

0.93

 recreation

0.93

 reconnaissance

0.92

 decoration

0.91

 souven

0.91

 protecting

0.89

Activations Density 0.422%

phrases related to reasons or purposes for actions

phrases related to consequences or purposes of actions

No Known Activations