INDEX

Explanations

names or entities with the suffix "-h"

oai_token-act-pair · gpt-3.5-turbo

the letter 'h' in various contexts

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

Not in Any Lists

No Comments

Negative Logits

-0.74

enhagen

-0.73

eering

-0.71

eers

-0.68

 flush

-0.67

 Weeks

-0.65

aution

-0.64

-0.61

 Clicker

-0.61

 Collider

-0.59

POSITIVE LOGITS

ythm

1.18

ulhu

1.18

orses

1.15

agen

1.13

atever

1.08

ollow

1.08

ouston

1.05

ope

0.99

ofer

0.99

onest

0.96

Activations Density 0.058%

names or entities with the suffix "-h"

the letter 'h' in various contexts

No Comments

No Known Activations