INDEX

Explanations

phrases indicating likelihood or speculation

oai_token-act-pair · gpt-3.5-turbo

statements that express assumptions or speculations about past events

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

opian

-0.67

essee

-0.66

ashi

-0.64

ackle

-0.64

ãĤŃ

-0.62

jri

-0.62

 Griff

-0.61

pour

-0.61

KI

-0.60

hetical

-0.58

POSITIVE LOGITS

 kidding

0.69

 gotten

0.68

 sensed

0.65

 liked

0.63

©¶æ

0.63

 metic

0.60

 wondered

0.60

ionics

0.60

 appreciated

0.59

 REALLY

0.59

Activations Density 0.087%

phrases indicating likelihood or speculation

statements that express assumptions or speculations about past events

No Comments

No Known Activations