İvme-Conversate-v2-Base (Codenamed Apple 2)

İvme (Turkish: acceleration) is a series of stupidly small language models built to punch above their weight. This is the second release: a 24M parameter decoder-only base model trained from scratch, this time with a much heavier training diet.

v1 produced grammatically correct sentences that did not really say anything when strung together, fluent without being about anything. v2 keeps almost the exact same architecture as v1 on purpose, and instead fixes the two things that actually mattered: far more training data, and a data mix weighted toward material that teaches a model to stay on topic across sentences.

---

Model Details

Nearly every setting above matches v1 on purpose. The point of v2 was to isolate the improvement to data and training, not to a bigger model.

---

Benchmarks

Benchmarks were run with lm-evaluation-harness via a custom model adapter (previous numbers from an internal script have been superseded by these).

BLiMP paradigm breakdown

Strong on core agreement paradigms:

Weaker on long-distance dependencies and island constraints, a known hard case for small models:

This pattern, strong local agreement paired with weaker long-distance syntax, is typical for models at this scale.

---

Does it actually make more sense now?

None of the benchmarks above directly test whether the model's writing holds together as connected text, which was v1's real problem. Sample output, EMA weights, temperature 0.8, top_k 50: Prompt: "Once upon a time, there was a"

Once upon a time, there was a wise old turtle named Timmy who lived on the coast of South America. In this magical place, no matter how big or large, people could look up at the sea and talk to each other. Timmy asked, "What do you mean, Timmy?" The turtle replied, "Well, I think you might see people talking about ocean creatures. They are like little waves that carry their voices. Sometimes they say they're too big or small to hear." After a few moments, Timmy had an idea. "Can we go on a boat-boat tour? I can't believe all the kids in the village are doing that!" As they sailed further, they saw many beautiful islands and vibrant colors. Each island had its unique culture and traditions. When they reached the top, they saw a group of kids playing and splashing around. "Wow, lookgies!" said Timmy. "They live in a big ocean full of colorful fish and

---

Training

Data Mix (~12.85B tokens, roughly 8 to 9x more than v1's 1.57B)

v1 was trained Chinchilla-optimal. v2 deliberately overtrains well past that point, since the model is small and cheap to run regardless.

Hyperparameters

Hardware

Trained on a single NVIDIA RTX PRO 6000 Blackwell (96GB) in approximately 4.75 hours.

---

Tokenizer

Custom byte-level BPE tokenizer trained from scratch on a sample of the pretraining mix. Vocab size 16,000.

---

Inference

This model can now be loaded with AutoModelForCausalLM instead of the manual pickle-loading workflow, and weights are available as model.safetensors.

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(
    "IvmeLabs/Ivme-Conversate-v2-Base", trust_remote_code=True, dtype=torch.float32,
)
tokenizer = AutoTokenizer.from_pretrained("IvmeLabs/Ivme-Conversate-v2-Base", trust_remote_code=True)
model.eval()
inputs = tokenizer("Once upon a time, there was a", return_tensors="pt")
out = model.generate(
    **inputs, max_new_tokens=200, do_sample=True,
    temperature=0.8, top_k=50, pad_token_id=tokenizer.pad_token_id,
)
print(tokenizer.decode(out[0], skip_special_tokens=True))

trust_remote_code=True is required (custom architecture: RoPE + SwiGLU + RMSNorm dense decoder). The original ckpt_final.pt pickle checkpoint and model/ architecture source remain in this repo unchanged for backwards compatibility.

Note on batch generation: use left-padding (tokenizer.padding_side = "left") the model doesn't use an explicit attention mask over padded positions, so right-padding within a batch will give incorrect results.

Legacy Inference

Here's a basic inference code in case you want to work with the pickle files.

import sys
import torch
from tokenizers import Tokenizer
from huggingface_hub import hf_hub_download, snapshot_download
repo_id = "IvmeLabs/Ivme-Conversate-v2-Base"
# Download just the model/ folder (architecture code) into the HF cache,
# then add it to sys.path so `from model import ...` works without the
# user needing to manually copy any files.
repo_local_dir = snapshot_download(repo_id, allow_patterns=["model/*"])
sys.path.append(repo_local_dir)
from model import IvmeConfig, IvmeConversateV2
ckpt_path = hf_hub_download(repo_id, "ckpt_final.pt")
tokenizer_path = hf_hub_download(repo_id, "tokenizer.json")
tokenizer = Tokenizer.from_file(tokenizer_path)
# IvmeConfig is a plain dataclass saved into the checkpoint. Trust this only
# because it's our own checkpoint, produced by our own training code.
torch.serialization.add_safe_globals([IvmeConfig])
ckpt = torch.load(ckpt_path, map_location="cuda")
cfg = ckpt["config"]
model = IvmeConversateV2(cfg)
# Use EMA weights (smoothed), not the raw training weights, for inference.
# Strip torch.compile's "_orig_mod." prefix if the checkpoint was compiled.
state_dict = ckpt["ema_state_dict"]
state_dict = {k.removeprefix("_orig_mod."): v for k, v in state_dict.items()}
model.load_state_dict(state_dict)
model.cuda().eval()
prompt = "Once upon a time, there was a"
ids = tokenizer.encode(prompt).ids
idx = torch.tensor([ids], dtype=torch.long, device="cuda")
eot_id = tokenizer.token_to_id("<|endoftext|>")
with torch.no_grad():
    for _ in range(200):
        idx_cond = idx if idx.size(1) <= cfg.context_len else idx[:, -cfg.context_len:]
        logits, _ = model(idx_cond)
        logits = logits[:, -1, :] / 0.8  # temperature
        v, _ = torch.topk(logits, 50)
        logits[logits < v[:, [-1]]] = -float("inf")
        probs = torch.softmax(logits, dim=-1)
        next_id = torch.multinomial(probs, num_samples=1)
        idx = torch.cat([idx, next_id], dim=1)
        if next_id.item() == eot_id:
            break
print(tokenizer.decode(idx[0].tolist()))

Limitations

• Base model only, not instruction tuned, will not follow instructions or answer questions
• English only
• 1024 token context window
• Weaker on long-distance syntactic dependencies than on local agreement, see BLiMP breakdown above
• No code data in the training mix (Python-Edu was dropped, see above)
• Weak knowledge. While the model has learned syntax and grammar very well, it falls short on knowledge and easily hallucinates. We are expecting to fix this in upcoming variants via distillation.

---

What's Next

Still on the table for a future version: distillation from a larger teacher model, and a return to the more experimental İvmetron architecture once time allows for the kind of patient debugging a genuinely novel design needs.

You can check our other upcoming models on our organization card!

---

Citation

@misc{ivme-conversate-v2-Base,
  author       = {IvmeLabs},
  title        = {İvme-Conversate-v2-Base},
  year         = {2026},
  publisher    = {Hugging Face},
  url          = {https://huggingface.co/IvmeLabs/Ivme-Conversate-v2-Base}
}

---

Credits

The apple photo by Matheus Cenali on Unsplash

---

Built by IvmeLabs. Small models, deliberate choices.