Chytrej2-90M-Base/README.md

language, license, pipeline_tag, tags, library_name, datasets

language	license	pipeline_tag	tags	library_name	datasets
en	apache-2.0	text-generation	llama causal-lm pretrained chytrej base	transformers	HuggingFaceFW/fineweb-edu

Chytrej2-90M-Base

A fully custom pretrained language model built from scratch on the LLaMA architecture trained on FineWeb Edu dataset.

Built by PingVortex Labs.

---

Model Details

• Parameters: 90M
• Context length: 8,192 tokens
• Language: English only
• Format: Base model
• Architecture: LLaMA
• License: Apache 2.0

---

Benchmarks

• No benchmarks - at this scale the benchmarks are more random than based on what the model has learned - we got a checkpoint 55k get better arc easy score than the final model.

---

Usage

from transformers import LlamaForCausalLM, PreTrainedTokenizerFast

model = LlamaForCausalLM.from_pretrained("pvlabs/Chytrej2-90M-Base")
tokenizer = PreTrainedTokenizerFast.from_pretrained("pvlabs/Chytrej2-90M-Base")

prompt = "The capital of France is"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=100, repetition_penalty=1.3)
print(tokenizer.decode(outputs[0]))

---

Made by PingVortex.