Upload folder using huggingface_hub
Browse files- configs/base.json +40 -0
- configs/nano.json +40 -0
configs/base.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"vocab_size": 16384,
|
| 4 |
+
"n_layers": 16,
|
| 5 |
+
"n_heads": 16,
|
| 6 |
+
"n_kv_heads": 4,
|
| 7 |
+
"d_model": 1024,
|
| 8 |
+
"d_ffn": 4096,
|
| 9 |
+
"max_seq_len": 1024,
|
| 10 |
+
"rope_theta": 10000.0,
|
| 11 |
+
"rms_eps": 1e-6,
|
| 12 |
+
"init_std": 0.02,
|
| 13 |
+
"dropout": 0.0,
|
| 14 |
+
"tie_embeddings": true,
|
| 15 |
+
"qk_norm": true,
|
| 16 |
+
"z_loss_coef": 1e-4
|
| 17 |
+
},
|
| 18 |
+
"tokenizer": {
|
| 19 |
+
"vocab_size": 16384,
|
| 20 |
+
"model_type": "bpe",
|
| 21 |
+
"character_coverage": 1.0,
|
| 22 |
+
"byte_fallback": true,
|
| 23 |
+
"normalization": "nmt_nfkc",
|
| 24 |
+
"split_digits": true,
|
| 25 |
+
"split_by_unicode_script": true,
|
| 26 |
+
"add_dummy_prefix": true,
|
| 27 |
+
"user_defined_symbols": [
|
| 28 |
+
"<|pad|>", "<|bos|>", "<|eos|>", "<|unk|>", "<|sep|>",
|
| 29 |
+
"<|system|>", "<|user|>", "<|assistant|>", "<|end|>",
|
| 30 |
+
"<|tool_call|>", "<|/tool_call|>", "<|tool_result|>", "<|/tool_result|>",
|
| 31 |
+
"<|cve|>", "<|cvss|>", "<|ioc|>", "<|ttp|>", "<|mitre|>", "<|kev|>",
|
| 32 |
+
"<|exploit|>", "<|patch|>", "<|alert|>",
|
| 33 |
+
"<|critical|>", "<|high|>", "<|medium|>", "<|low|>", "<|info|>"
|
| 34 |
+
],
|
| 35 |
+
"balance": {
|
| 36 |
+
"conversational_ratio": 0.5,
|
| 37 |
+
"technical_ratio": 0.5
|
| 38 |
+
}
|
| 39 |
+
}
|
| 40 |
+
}
|
configs/nano.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"vocab_size": 16384,
|
| 4 |
+
"n_layers": 8,
|
| 5 |
+
"n_heads": 8,
|
| 6 |
+
"n_kv_heads": 2,
|
| 7 |
+
"d_model": 512,
|
| 8 |
+
"d_ffn": 2304,
|
| 9 |
+
"max_seq_len": 1024,
|
| 10 |
+
"rope_theta": 10000.0,
|
| 11 |
+
"rms_eps": 1e-6,
|
| 12 |
+
"init_std": 0.02,
|
| 13 |
+
"dropout": 0.0,
|
| 14 |
+
"tie_embeddings": true,
|
| 15 |
+
"qk_norm": true,
|
| 16 |
+
"z_loss_coef": 1e-4
|
| 17 |
+
},
|
| 18 |
+
"tokenizer": {
|
| 19 |
+
"vocab_size": 16384,
|
| 20 |
+
"model_type": "bpe",
|
| 21 |
+
"character_coverage": 1.0,
|
| 22 |
+
"byte_fallback": true,
|
| 23 |
+
"normalization": "nmt_nfkc",
|
| 24 |
+
"split_digits": true,
|
| 25 |
+
"split_by_unicode_script": true,
|
| 26 |
+
"add_dummy_prefix": true,
|
| 27 |
+
"user_defined_symbols": [
|
| 28 |
+
"<|pad|>", "<|bos|>", "<|eos|>", "<|unk|>", "<|sep|>",
|
| 29 |
+
"<|system|>", "<|user|>", "<|assistant|>", "<|end|>",
|
| 30 |
+
"<|tool_call|>", "<|/tool_call|>", "<|tool_result|>", "<|/tool_result|>",
|
| 31 |
+
"<|cve|>", "<|cvss|>", "<|ioc|>", "<|ttp|>", "<|mitre|>", "<|kev|>",
|
| 32 |
+
"<|exploit|>", "<|patch|>", "<|alert|>",
|
| 33 |
+
"<|critical|>", "<|high|>", "<|medium|>", "<|low|>", "<|info|>"
|
| 34 |
+
],
|
| 35 |
+
"balance": {
|
| 36 |
+
"conversational_ratio": 0.5,
|
| 37 |
+
"technical_ratio": 0.5
|
| 38 |
+
}
|
| 39 |
+
}
|
| 40 |
+
}
|