INDEX

Explanations

references to urgent requests or appeals made in various contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "call" in various contexts related to requests or appeals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Fiction

-0.71

 concentration

-0.71

 ferment

-0.65

 Flavoring

-0.62

izontal

-0.61

theless

-0.61

poon

-0.61

 incorpor

-0.60

 Scotia

-0.59

 aquarium

-0.59

POSITIVE LOGITS

backs

1.27

igraph

1.12

call

1.11

oused

0.90

 caller

0.80

outs

0.80

calling

0.79

ers

0.78

cards

0.77

tower

0.77

Activations Density 0.034%

references to urgent requests or appeals made in various contexts

instances of the word "call" in various contexts related to requests or appeals

No Comments

No Known Activations