INDEX

Explanations

the word "guess" followed by a statement/question

oai_token-act-pair · gpt-3.5-turbo

instances of the word "guess" and related phrases indicating uncertainty or making inferences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

interrupted

-0.85

iences

-0.77

contained

-0.69

elight

-0.69

iterranean

-0.67

ĸļ

-0.66

tha

-0.65

obe

-0.65

ksh

-0.65

helle

-0.64

POSITIVE LOGITS

 guesses

1.14

 guess

0.93

 guessed

0.80

 guessing

0.80

 Guess

0.75

lessly

0.72

itial

0.70

IELD

0.67

lege

0.67

 incorrectly

0.66

Activations Density 0.026%

the word "guess" followed by a statement/question

instances of the word "guess" and related phrases indicating uncertainty or making inferences

No Comments

No Known Activations