INDEX

Explanations

specific names or words related to locations or activities

oai_token-act-pair · gpt-3.5-turbo

occurrences of the character sequence "ch"

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

 fuse

-0.66

ACTED

-0.64

 flourish

-0.58

 sanity

-0.58

 whistle

-0.57

ATHER

-0.57

 accuracy

-0.56

 reconc

-0.56

 pointer

-0.56

AVG

-0.55

POSITIVE LOGITS

imura

0.88

ilon

0.87

ula

0.87

ivas

0.85

awar

0.84

ima

0.83

ulas

0.82

ara

0.82

ter

0.81

illes

0.80

Activations Density 0.106%

specific names or words related to locations or activities

occurrences of the character sequence "ch"

No Comments

No Known Activations