INDEX

Explanations

phrases related to continuous or persistent actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "stop" and its variations, signaling a focus on continuity or interruptions

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

axter

-0.71

Osw

-0.70

imar

-0.68

 condem

-0.66

icho

-0.63

okia

-0.62

ischer

-0.62

rote

-0.62

ahime

-0.62

 Clerk

-0.61

POSITIVE LOGITS

theless

0.83

length

0.74

love

0.74

etheless

0.73

aneously

0.72

mouth

0.68

ency

0.68

ances

0.68

motion

0.68

withstanding

0.66

Activations Density 0.027%

phrases related to continuous or persistent actions

instances of the word "stop" and its variations, signaling a focus on continuity or interruptions

No Comments

No Known Activations