Update model weights
Browse files- config.json +34 -0
- model.safetensors +3 -0
config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"MamLAForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"d_model": 2048,
|
| 7 |
+
"dtype": "float32",
|
| 8 |
+
"eos_token_id": 2,
|
| 9 |
+
"mamba_d_conv": 4,
|
| 10 |
+
"mamba_d_state": 64,
|
| 11 |
+
"mamba_expand": 2,
|
| 12 |
+
"mamba_layers_per_cycle": 7,
|
| 13 |
+
"mamba_n_heads": 16,
|
| 14 |
+
"mla_kv_lora_rank": 128,
|
| 15 |
+
"mla_layers_per_cycle": 1,
|
| 16 |
+
"mla_n_heads": 16,
|
| 17 |
+
"mla_q_lora_rank": 128,
|
| 18 |
+
"mla_qk_nope_head_dim": 32,
|
| 19 |
+
"mla_qk_rope_head_dim": 32,
|
| 20 |
+
"mla_v_head_dim": 32,
|
| 21 |
+
"model_type": "mamla",
|
| 22 |
+
"moe_d_ff": 1024,
|
| 23 |
+
"moe_num_experts": 4,
|
| 24 |
+
"moe_top_k": 2,
|
| 25 |
+
"n_layer": 24,
|
| 26 |
+
"num_experts": 4,
|
| 27 |
+
"num_experts_per_tok": 2,
|
| 28 |
+
"num_layers": 16,
|
| 29 |
+
"pad_token_id": 0,
|
| 30 |
+
"pad_vocab_size_multiple": 8,
|
| 31 |
+
"tie_word_embeddings": true,
|
| 32 |
+
"transformers_version": "5.0.0",
|
| 33 |
+
"vocab_size": 32000
|
| 34 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:41a4536def960e98b77a5818b2ca97b5e19b506c4aced51389a887c320f106bd
|
| 3 |
+
size 3602406968
|