INDEX

Explanations

the word "proved."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "proved" 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 @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 newsletters

-0.80

 contrace

-0.64

 throats

-0.64

 mouth

-0.63

 lockout

-0.61

atched

-0.61

alez

-0.61

 hazard

-0.61

 bios

-0.59

letal

-0.59

POSITIVE LOGITS

\\\\\\\\

0.89

 proves

0.82

ãĤ¤ãĥĪ

0.79

 proved

0.78

ãĥ¼ãĥĨ

0.78

ingen

0.74

 dispro

0.72

nces

0.71

GGGGGGGG

0.70

»

0.69

Activations Density 0.012%

the word "proved."

instances of the word "proved" in various contexts

No Comments

No Known Activations