Cooldown / annealing run initialized from Kestrel-34M-base with a mix including 20% synthetic step-by-step arithmetic.
prelude=1, core=4 x 3 loops, coda=1, hidden_size=512, attention_window=256, memory_chunk_size=64, memory_heads=4, vocab_size=16384)This model uses a custom KestrelForCausalLM architecture (trust_remote_code=True):
from transformers import AutoModelForCausalLM, AutoTokenizer
repo_id = "lillycham/Kestrel-34M-anneal-base"
tokenizer = AutoTokenizer.from_pretrained(repo_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(repo_id, trust_remote_code=True)
prompt = tokenizer.apply_chat_template([{"role": "user", "content": "Hello!"}], tokenize=False, add_generation_prompt=True)
inputs = tokenizer(prompt, return_tensors="pt")
out = model.generate(inputs.input_ids, max_new_tokens=64, temperature=0.7, top_p=0.9)
print(tokenizer.decode(out[0], skip_special_tokens=False))
5000, Batch size: 8 x grad_accum 4, Sequence length: 1024lr=0.015) + AdamW (lr=0.00075), weight decay 0.01bf16 on cuda