INDEX

Explanations

words related to correctness, accuracy, or being in the right situation or position

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "right" in various contexts

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

ĸļ

-0.77

odium

-0.73

DAQ

-0.71

ulative

-0.70

mat

-0.68

ains

-0.68

 srfAttach

-0.67

 Clarke

-0.65

 Peaks

-0.64

è¦ļéĨĴ

-0.63

POSITIVE LOGITS

eous

1.07

 wing

0.92

shore

0.85

terday

0.82

wing

0.78

 forearm

0.77

ward

0.75

 fielder

0.73

lihood

0.73

lander

0.72

Activations Density 0.055%

words related to correctness, accuracy, or being in the right situation or position

mentions of the word "right" in various contexts

No Comments

No Known Activations