INDEX

Explanations

names of specific individuals, likely related to news or events

oai_token-act-pair · gpt-3.5-turbo

particular strings of capital letters and specific punctuation or formatting cues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Bel

-0.85

 Arnold

-0.85

Lod

-0.79

 Lewis

-0.79

 Dwell

-0.79

 Undead

-0.78

stal

-0.77

 Shelley

-0.76

 Lords

-0.75

Ð²

-0.75

POSITIVE LOGITS

ap

1.54

AP

1.46

aps

1.46

APS

1.33

ip

1.32

apping

1.25

EP

1.24

mp

1.21

apped

1.17

apo

1.16

Activations Density 0.271%

names of specific individuals, likely related to news or events

particular strings of capital letters and specific punctuation or formatting cues

No Comments

No Known Activations