INDEX

Explanations

words related to telegrams

oai_token-act-pair · gpt-3.5-turbo

mentions of telegrams and related content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 overrun

-0.66

Ai

-0.65

 immune

-0.65

 Occup

-0.64

Oo

-0.62

pop

-0.62

Mis

-0.61

 biome

-0.61

CLS

-0.60

 Aber

-0.59

POSITIVE LOGITS

legram

4.40

 Telegram

2.36

legraph

2.13

astern

1.49

hran

1.27

imeo

1.10

alions

1.06

ewitness

1.04

 Lama

1.04

cester

0.97

Activations Density 0.023%

words related to telegrams

mentions of telegrams and related content

No Comments

No Known Activations