Echo88-150M-Base

Echo88-150M-Base is a small English decoder-only causal language model trained from scratch on the Echo88 pretraining dataset.

Echo88 is designed as a retro language model inspired by the language, culture, computing, magazines, Usenet discussions, and older book text available up to the late 1980s.

This is a base model, not an instruction-tuned chatbot. It is trained for next-token prediction and should be fine-tuned before being used as a helpful assistant.

Model Details

• Model name: Echo88-150M-Base
• Model type: decoder-only causal language model
• Architecture: LLaMA-style transformer
• Training type: from scratch
• Parameter count: 163,606,272 parameters
• Language: English
• Context length: 2048 tokens
• Tokenizer: custom Echo88 byte-level BPE tokenizer
• Vocabulary size: 32,768
• Training objective: autoregressive next-token prediction

Training Data

Echo88-150M-Base was trained on the Echo88 pretraining dataset.

The packed training set contains:

• Train tokens: 1,470,629,888
• Eval tokens: 1,454,080
• Train blocks: 718,081 blocks
• Eval blocks: 710 blocks
• Block size: 2048 tokens
• Packed dtype: uint16

The dataset includes a mixture of:

• public-domain book text
• Gutenberg-style older books
• UTZOO Usenet posts
• BYTE Magazine text
• PC Magazine text
• TIME Magazine text
• Internet Archive Magazine Rack OCR text
• computer and technology magazine text
• general historical magazine text

The dataset emphasizes the 1950s through the late 1980s, with a strong focus on early personal computing, printed magazines, Usenet, and older long-form writing.

Dataset used:

• guus4324343/Echo88-Pretrain-1.17B

Intended Use

Echo88-150M-Base is intended for:

• causal language modeling
• retro / historical AI experiments
• small language model research
• continued pretraining
• instruction tuning
• 1980s-style assistant experiments
• computer-history language modeling
• training Echo88-150M-Instruct

Recommended flow:

Echo88-150M-Base
→ supervised fine-tuning on Echo88-Instruct-173K
→ Echo88-150M-Instruct

Not Instruction Tuned

This model is not instruction tuned.

It may not reliably follow commands, answer questions directly, or behave like a chat assistant. It is a base model that continues text.

Expected behavior:

• continues prompts
• completes paragraphs
• imitates old magazine/book/Usenet style
• may produce raw text instead of direct answers
• may hallucinate
• may repeat phrases
• may generate OCR-like artifacts

For chat behavior, use or create an instruction-tuned version using:

• guus4324343/Echo88-Instruct-173K

Knowledge Boundary

Echo88 is designed around a historical data mixture ending around the late 1980s.

The model should not be expected to know modern topics such as:

• Google
• Wikipedia
• iPhone
• smartphones
• modern social media
• Windows 95 and later software
• COVID-19
• modern AI systems
• 2000s, 2010s, or 2020s events

Because this is a base model, it may still hallucinate if prompted about modern events. The later instruction-tuned model should be trained to respond more carefully to post-1988 topics.

Example Usage

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

model_id = "guus4324343/Echo88-150M-Base"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

prompt = "The personal computer revolution of the 1980s"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)

with torch.no_grad():
    output = model.generate(
        **inputs,
        max_new_tokens=160,
        temperature=0.8,
        top_p=0.95,
        do_sample=True,
        repetition_penalty=1.05,
        pad_token_id=tokenizer.pad_token_id,
        eos_token_id=tokenizer.eos_token_id,
    )

print(tokenizer.decode(output[0], skip_special_tokens=True))

Training Configuration

Echo88-150M-Base was trained as a LLaMA-style decoder-only causal LM.

Main configuration:

vocab_size: 32768
hidden_size: 768
intermediate_size: 2048
num_hidden_layers: 18
num_attention_heads: 12
num_key_value_heads: 4
max_position_embeddings: 2048
activation: SiLU / SwiGLU-style LLaMA MLP
normalization: RMSNorm
position encoding: RoPE
attention: grouped-query attention

Training setup:

precision: bf16
sequence length: 2048
optimizer: AdamW
scheduler: cosine
weight decay: 0.1
gradient clipping: 1.0
max steps: 5610
training tokens: ~1.47B

Limitations

Echo88-150M-Base is experimental and small.

Known limitations:

• not instruction tuned
• may hallucinate
• may repeat text
• may produce OCR-like artifacts
• may reflect outdated historical language or views
• may struggle with complex reasoning
• may not reliably refuse post-1988 topics
• may produce incomplete or strange continuations
• may mix unrelated historical/computer facts

The model is intended for research, experimentation, and creative retro AI work. It is not intended for high-stakes use.

Bias and Historical Content

The training data includes historical books, magazines, and Usenet text. As a result, the model may reproduce outdated language, assumptions, stereotypes, or viewpoints present in older source material.

Users should review outputs carefully.

Model Family

Planned Echo88 model family:

Echo88-150M-Base
Echo88-150M-Instruct
Echo88-150M-Chat

License

The model weights are released under the Apache 2.0 license.

The training dataset is mixed-source and released separately under other. Users are responsible for checking dataset source rights, licensing, and suitability for their own use case.