INDEX

Explanations

phrases related to responses or reactions from individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "responded" and its variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

borne

-0.78

tan

-0.76

wed

-0.75

cgi

-0.72

dar

-0.71

free

-0.71

holding

-0.71

fo

-0.71

bons

-0.71

lore

-0.69

POSITIVE LOGITS

 angrily

0.96

 harshly

0.85

 sarcast

0.84

 favorably

0.84

 thereto

0.78

 indign

0.75

 accordingly

0.72

isson

0.70

 explan

0.70

reply

0.70

Activations Density 0.026%

phrases related to responses or reactions from individuals

instances of the word "responded" and its variations

No Comments

No Known Activations