INDEX

Explanations

words related to specific geographical locations, with a focus on California

oai_token-act-pair · gpt-3.5-turbo

references to the state of California and variations of the name Longhorns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

theless

-0.89

tsky

-0.77

anwhile

-0.71

ueller

-0.65

 combustion

-0.63

bed

-0.63

 Loving

-0.62

 guidance

-0.61

 fetal

-0.60

isen

-0.60

POSITIVE LOGITS

ians

0.97

esse

0.88

enos

0.87

 citiz

0.85

arians

0.82

 Hots

0.81

esc

0.79

ian

0.77

oslov

0.76

escent

0.76

Activations Density 0.012%

words related to specific geographical locations, with a focus on California

references to the state of California and variations of the name Longhorns

No Comments

No Known Activations