INDEX

Explanations

adjectives or verbs related to being impressed

oai_token-act-pair · gpt-3.5-turbo

expressions of admiration or impact

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 hijacked

-0.66

adobe

-0.66

nia

-0.64

space

-0.64

 sovere

-0.62

angled

-0.60

 saline

-0.60

umble

-0.59

 starch

-0.58

yll

-0.58

POSITIVE LOGITS

mented

1.09

ments

1.06

ively

1.02

ment

0.93

mentation

0.91

MENT

0.84

iveness

0.84

ences

0.83

enced

0.83

encer

0.81

Activations Density 0.062%

adjectives or verbs related to being impressed

expressions of admiration or impact

No Comments

No Known Activations