mirror of
https://github.com/zenlm/zen-family.git
synced 2026-11-03 14:26:45 +00:00
72 lines
1.8 KiB
JSON
72 lines
1.8 KiB
JSON
{
|
|
"model_name": "zen-nano-instruct",
|
|
"metadata": {
|
|
"model_name": "zen-nano-instruct",
|
|
"model_type": "instruction-following",
|
|
"version": "1.0.0",
|
|
"has_thinking": false,
|
|
"special_tokens": {
|
|
"<s>": "bos_token",
|
|
"</s>": "eos_token",
|
|
"<pad>": "pad_token",
|
|
"<|user|>": "user_token",
|
|
"<|assistant|>": "assistant_token"
|
|
},
|
|
"context_length": 8192,
|
|
"vocabulary_size": 32000,
|
|
"hidden_size": 768,
|
|
"num_layers": 12,
|
|
"num_heads": 12,
|
|
"author": "Hanzo AI",
|
|
"license": "Apache-2.0",
|
|
"base_model": null,
|
|
"capabilities": [
|
|
"instruction-following",
|
|
"chat",
|
|
"coding"
|
|
],
|
|
"training_data": "Hanzo proprietary dataset",
|
|
"quantization_info": null
|
|
},
|
|
"conversion_settings": {
|
|
"context_size": 8192,
|
|
"vocab_type": "bpe",
|
|
"use_f16": true,
|
|
"quantization_options": {
|
|
"mobile": [
|
|
"Q4_K_S",
|
|
"Q4_K_M"
|
|
],
|
|
"balanced": [
|
|
"Q5_K_M",
|
|
"Q4_K_M"
|
|
],
|
|
"quality": [
|
|
"Q6_K",
|
|
"Q8_0"
|
|
],
|
|
"server": [
|
|
"Q8_0",
|
|
"FP16"
|
|
]
|
|
}
|
|
},
|
|
"gguf_metadata": {
|
|
"general.name": "zen-nano-instruct",
|
|
"general.architecture": "instruction_following",
|
|
"general.author": "Hanzo AI",
|
|
"general.version": "1.0.0",
|
|
"general.license": "Apache-2.0",
|
|
"general.description": "Zen instruction-following model by Hanzo AI",
|
|
"instruction-following.context_length": 8192,
|
|
"instruction-following.embedding_length": 768,
|
|
"instruction-following.block_count": 12,
|
|
"instruction-following.attention.head_count": 12,
|
|
"instruction-following.vocabulary_size": 32000,
|
|
"tokenizer.ggml.model": "gpt2",
|
|
"tokenizer.ggml.bos_token_id": 1,
|
|
"tokenizer.ggml.eos_token_id": 2,
|
|
"tokenizer.ggml.padding_token_id": 0,
|
|
"tokenizer.ggml.unknown_token_id": 3
|
|
}
|
|
} |