INDEX

Explanations

terms related to intense focus or concentration

oai_token-act-pair · gpt-3.5-turbo

occurrences of the substring "int" 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 @ 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

hyde

-0.90

wagen

-0.78

ĸļ

-0.77

Reviewer

-0.76

buster

-0.73

Train

-0.69

jriwal

-0.68

SEA

-0.66

SHIP

-0.66

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.66

POSITIVE LOGITS

ention

1.09

ellect

1.06

ensions

1.00

itial

0.98

ension

0.94

ended

0.93

ract

0.92

icult

0.92

ensive

0.90

ensity

0.89

Activations Density 0.014%

terms related to intense focus or concentration

occurrences of the substring "int" in various contexts

No Comments

No Known Activations