Downloads · 30 days
1
6% of all-time downloads
dreamer1cc/hovor-eurollm-1.7b-coreml-int4
hovor-eurollm-1.7b-coreml-int4 is a machine learning model from dreamer1cc. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
CoreML conversion of utter-project/EuroLLM-1.7B-Instruct for Apple Neural Engine (iOS 18+ / macOS 15+).
Downloads · 30 days
1
6% of all-time downloads
All-time downloads
16
Public
Repo size
833 MB
Likes
0
Public
Click a slice to open those files.
.bin829 MB · 99%
From the Hugging Face model README
CoreML conversion of utter-project/EuroLLM-1.7B-Instruct for Apple Neural Engine (iOS 18+ / macOS 15+).
Apache-2.0 (inherited from source model)
| Parameter | Value |
|---|---|
| Source | utter-project/EuroLLM-1.7B-Instruct |
| Architecture | LlamaForCausalLM (llama) |
| Quantization | int4 palettization (group_size=32) |
| Context length | 512 |
| Compute units | CPU_AND_NE (ANE-safe, 0 GPU-pinned ops) |
| Converter | john-rocky/CoreML-LLM (MonolithicWrapper) |
| coremltools | 8.3 |
verify_ane.py result: 5152 ops inspected, 0 GPU-preferred ops — PASS
{
"architecture": "llama",
"hidden_size": 2048,
"num_hidden_layers": 24,
"num_attention_heads": 16,
"num_key_value_heads": 8,
"head_dim": 128,
"vocab_size": 128000,
"context_length": 512,
"rms_norm_eps": 1e-5,
"quantization": "int4",
"compute_units": "ALL"
}
EuroLLM uses ChatML format — NOT standard Llama-3 turn markers.
<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{user_message}<|im_end|>
<|im_start|>assistant
<s><|im_end|> (token id 4)LlamaTokenizerRuntime note: If the CoreML runtime uses arch "llama" to select a buildPrompt function, verify it applies ChatML markers (<|im_start|> / <|im_end|>) rather than Llama-3 markers (<|begin_of_text|> / <|eot_id|>). The two are incompatible — using Llama-3 markers on EuroLLM will produce garbage output.
model.mlpackage — monolithic CoreML model (embed + transformer + lm_head, stateful KV cache)model_config.json — metadata for Swift inference enginehf_model/ — downloaded HF snapshot (tokenizer files; model.safetensors excluded as redundant)