INDEX

Explanations

actions, events, and processes related to conflict and resolution

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

yond

-0.77

tis

-0.71

wcsstore

-0.68

arta

-0.68

wayne

-0.66

etric

-0.66

Present

-0.65

ija

-0.62

etrical

-0.62

yah

-0.62

POSITIVE LOGITS

 extensively

0.95

 unsuccessfully

0.94

nesday

0.86

 yesterday

0.84

 briefly

0.81

 unanimously

0.80

him

0.78

 tremend

0.74

 last

0.72

 LAST

0.69

Activations Density 4.811%

actions, events, and processes related to conflict and resolution

No Comments

No Known Activations