INDEX

Explanations

names of specific individuals or places

oai_token-act-pair · gpt-3.5-turbo

proper names and references to unconventional or indiscriminate actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

keeping

-0.81

âĶģ

-0.80

hyde

-0.75

keeper

-0.73

 Redemption

-0.73

Problem

-0.72

stores

-0.72

omatic

-0.70

Benef

-0.69

keepers

-0.69

POSITIVE LOGITS

inately

0.90

ishable

0.86

nel

0.86

ensibly

0.82

ifer

0.76

rative

0.74

ently

0.72

mble

0.72

elve

0.72

vre

0.72

Activations Density 0.030%

names of specific individuals or places

proper names and references to unconventional or indiscriminate actions

No Comments

No Known Activations