INDEX

Explanations

words related to evasive actions or decisions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the letter "d" 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 @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

MENTS

-0.61

men

-0.58

MAX

-0.57

 latex

-0.57

 feminine

-0.57

MENT

-0.56

 female

-0.56

End

-0.55

(>

-0.55

HBO

-0.55

POSITIVE LOGITS

umbered

1.27

owed

1.20

ucked

1.20

umped

1.13

ussed

1.11

inged

1.08

itched

1.07

rolled

1.06

ounced

1.05

uked

1.04

Activations Density 0.068%

words related to evasive actions or decisions

occurrences of the letter "d" in various contexts

No Known Activations