INDEX

Explanations

adjectives with a negative or challenging connotation

oai_token-act-pair · gpt-3.5-turbo

statements indicating conditions or situations that are ongoing, repeated, or have been established over time

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ennett

-0.76

midt

-0.68

ials

-0.66

rison

-0.66

ants

-0.65

addon

-0.62

hov

-0.60

 Serpent

-0.60

'>

-0.60

manager

-0.60

POSITIVE LOGITS

 incidentally

0.64

 genetically

0.63

pport

0.63

 antit

0.61

 inhabit

0.61

 incompatible

0.61

pires

0.60

 relegation

0.59

fit

0.59

 inconvenient

0.58

Activations Density 0.830%

adjectives with a negative or challenging connotation

statements indicating conditions or situations that are ongoing, repeated, or have been established over time

No Comments

No Known Activations