ZERO LOOP base weights
Browse files- base/chat_template.jinja +1 -0
- base/config.json +51 -0
- base/generation_config.json +9 -0
- base/model.safetensors +3 -0
- base/tokenizer.json +0 -0
- base/tokenizer_config.json +8 -0
base/chat_template.jinja
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{% if messages[0]['role'] != 'system' %}<|system|>You are Zero Loop, an AI assistant created by Zero Labs. Your name is Zero Loop and your creator is Zero Labs. Think inside <think></think> before you answer. Tools: search (argument query) looks up Wikipedia, calculator (argument expression) evaluates arithmetic. To call a tool output <tool_call> then a JSON object with keys name and arguments then </tool_call>, and wait for the tool result.<|end|>{% endif %}{% for m in messages %}<|{{ m['role'] }}|>{{ m['content'] }}<|end|>{% endfor %}{% if add_generation_prompt %}<|assistant|>{{ '<think>\n' }}{% endif %}
|
base/config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"Qwen3ForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": null,
|
| 8 |
+
"dtype": "float32",
|
| 9 |
+
"eos_token_id": 16379,
|
| 10 |
+
"head_dim": 64,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 256,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 1024,
|
| 15 |
+
"layer_types": [
|
| 16 |
+
"full_attention",
|
| 17 |
+
"full_attention",
|
| 18 |
+
"full_attention",
|
| 19 |
+
"full_attention",
|
| 20 |
+
"full_attention",
|
| 21 |
+
"full_attention",
|
| 22 |
+
"full_attention",
|
| 23 |
+
"full_attention",
|
| 24 |
+
"full_attention",
|
| 25 |
+
"full_attention",
|
| 26 |
+
"full_attention",
|
| 27 |
+
"full_attention",
|
| 28 |
+
"full_attention",
|
| 29 |
+
"full_attention",
|
| 30 |
+
"full_attention",
|
| 31 |
+
"full_attention"
|
| 32 |
+
],
|
| 33 |
+
"max_position_embeddings": 4096,
|
| 34 |
+
"max_window_layers": 28,
|
| 35 |
+
"model_type": "qwen3",
|
| 36 |
+
"num_attention_heads": 4,
|
| 37 |
+
"num_hidden_layers": 16,
|
| 38 |
+
"num_key_value_heads": 2,
|
| 39 |
+
"pad_token_id": 16374,
|
| 40 |
+
"rms_norm_eps": 1e-06,
|
| 41 |
+
"rope_parameters": {
|
| 42 |
+
"rope_theta": 10000.0,
|
| 43 |
+
"rope_type": "default"
|
| 44 |
+
},
|
| 45 |
+
"sliding_window": null,
|
| 46 |
+
"tie_word_embeddings": true,
|
| 47 |
+
"transformers_version": "5.18.0",
|
| 48 |
+
"use_cache": true,
|
| 49 |
+
"use_sliding_window": false,
|
| 50 |
+
"vocab_size": 16384
|
| 51 |
+
}
|
base/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"eos_token_id": 16379,
|
| 4 |
+
"output_attentions": false,
|
| 5 |
+
"output_hidden_states": false,
|
| 6 |
+
"pad_token_id": 16374,
|
| 7 |
+
"transformers_version": "5.18.0",
|
| 8 |
+
"use_cache": true
|
| 9 |
+
}
|
base/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fbf4697178084eecc9094da0750e706bb6c95dc01effcafe0e7087493f1662f5
|
| 3 |
+
size 79753104
|
base/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
base/tokenizer_config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"clean_up_tokenization_spaces": false,
|
| 4 |
+
"eos_token": "<|end|>",
|
| 5 |
+
"model_max_length": 4096,
|
| 6 |
+
"pad_token": "<|pad|>",
|
| 7 |
+
"tokenizer_class": "TokenizersBackend"
|
| 8 |
+
}
|