GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Convergent-ch
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
Convergent-ch
Model Tree
Input Modalities
Output Modalities
Supported Functionality
| Parameters | 254,400 |
| Architecture | Llama-style decoder (RMSNorm, SwiGLU, RoPE, tied embeddings) |
| Layers / hidden / MLP | 3 / 64 / 128 |
| Attention heads | 2 (head dim 32) |
| Context | 512 tokens |
| Vocabulary | 2,048 (byte-level BPE, trained on the chat data) |
| Language | English |
model.safetensors, config.json, tokenizer files: Hugging Face transformers format (LlamaForCausalLM).pico1.gguf: GGUF (f16) for Ollama and llama.cpp.Modelfile: Ollama model file with the chat format and stop tokens.Download pico1.gguf and Modelfile into one folder, then:
bash
ollama create pico1 -f Modelfileollama run pico1python
from transformers import AutoTokenizer, AutoModelForCausalLM tok = AutoTokenizer.from_pretrained("Convergent-ch/Pico-1")model = AutoModelForCausalLM.from_pretrained("Convergent-ch/Pico-1") ids = tok.apply_chat_template([{"role": "user", "content": "What is your name?"}], add_generation_prompt=True, return_tensors="pt")out = model.generate(ids, max_new_tokens=40, do_sample=False)print(tok.decode(out[0, ids.shape[1]:], skip_special_tokens=True))# My name's Pico 1. I'm a tiny AI made by Eron.text
<|bos|><|user|>Hello!<|end|><|assistant|>Generation stops at <|end|>. Low temperatures (0 to 0.3) work best.
Trained from scratch on about 19.8M tokens of short English chats (about 108,700 conversations), 18,000 steps with batch size 32 x 512 tokens (about 15 epochs), AdamW, learning rate 3e-3 with cosine decay, bf16. Training took about 6 minutes on one RTX 3060 Ti. Final validation loss 3.00 (perplexity 20.1).
Training data:
Apache 2.0.
ollama create pico1 -f Modelfileollama run pico1from transformers import AutoTokenizer, AutoModelForCausalLM tok = AutoTokenizer.from_pretrained("Convergent-ch/Pico-1")model = AutoModelForCausalLM.from_pretrained("Convergent-ch/Pico-1") ids = tok.apply_chat_template([{"role": "user", "content": "What is your name?"}], add_generation_prompt=True, return_tensors="pt")out = model.generate(ids, max_new_tokens=40, do_sample=False)print(tok.decode(out[0, ids.shape[1]:], skip_special_tokens=True))# My name's Pico 1. I'm a tiny AI made by Eron.