INDEX

Explanations

phrases related to controversies or serious accusations

oai_token-act-pair · gpt-3.5-turbo

phrases related to references or descriptions of actions and concepts involving "of."

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

casting

-0.69

bet

-0.67

 preval

-0.63

 istg

-0.61

 acronym

-0.61

ringe

-0.60

monitor

-0.60

tenance

-0.60

 projector

-0.60

flix

-0.59

POSITIVE LOGITS

ãĤ¯

0.66

 sorts

0.65

ãĥĺ

0.64

irgin

0.64

 course

0.63

 minorities

0.62

glers

0.61

 innocent

0.61

odan

0.61

 certain

0.60

Activations Density 0.194%

phrases related to controversies or serious accusations

phrases related to references or descriptions of actions and concepts involving "of."

No Comments

No Known Activations