INDEX

Explanations

adjectives or phrases related to negativity or aggression

oai_token-act-pair · gpt-3.5-turbo

instances of the word "hostile" 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 @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

akings

-0.85

ucket

-0.84

ingham

-0.80

ulet

-0.80

oled

-0.77

bind

-0.77

orah

-0.77

-0.76

agate

-0.76

andre

-0.75

POSITIVE LOGITS

 hostile

1.06

 takeover

0.96

 undermin

0.84

 hostility

0.84

 citiz

0.82

 dismissal

0.80

 retaliation

0.80

 invaders

0.76

 invasion

0.76

 interference

0.73

Activations Density 0.010%

adjectives or phrases related to negativity or aggression

instances of the word "hostile" in various contexts

No Comments

No Known Activations