INDEX

Explanations

adjectives related to a strong positive trait or evaluation

oai_token-act-pair · gpt-3.5-turbo

concepts related to reputation and influence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Correction

-0.68

HQ

-0.67

USS

-0.67

 Situation

-0.65

COP

-0.65

Activity

-0.65

ENTION

-0.64

kar

-0.64

RD

-0.64

 Discussion

-0.63

POSITIVE LOGITS

 outweigh

1.48

 outwe

1.36

 evapor

1.15

 dwind

1.14

 overshadow

1.06

 trump

1.04

 perme

1.01

 waning

1.00

 dissip

0.98

 manifested

0.98

Activations Density 0.262%

adjectives related to a strong positive trait or evaluation

concepts related to reputation and influence

No Comments

No Known Activations