Checkpoint checkpoints/1388660k (step 64,737)
Browse files
checkpoints/1388660k/config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"name": "nirca-mini",
|
| 3 |
+
"step": 64737,
|
| 4 |
+
"tokens": 1388660099,
|
| 5 |
+
"quant": "ternary",
|
| 6 |
+
"quant_scope": "all",
|
| 7 |
+
"n_experts": 32,
|
| 8 |
+
"chat_format": "chatml",
|
| 9 |
+
"bytes": 32638048,
|
| 10 |
+
"cfg": {
|
| 11 |
+
"vocab_size": 265,
|
| 12 |
+
"d_model": 512,
|
| 13 |
+
"n_head": 8,
|
| 14 |
+
"d_ff": 1408,
|
| 15 |
+
"block": 2048,
|
| 16 |
+
"rope_theta": 10000.0,
|
| 17 |
+
"tie_embeddings": true,
|
| 18 |
+
"n_prelude": 2,
|
| 19 |
+
"n_recur": 1,
|
| 20 |
+
"n_coda": 0,
|
| 21 |
+
"max_steps": 24,
|
| 22 |
+
"pool_experts": 32,
|
| 23 |
+
"pool_d_ff": 2048,
|
| 24 |
+
"pool_top_k": 8,
|
| 25 |
+
"pool_resident": 32,
|
| 26 |
+
"pool_capacity_factor": 2.25,
|
| 27 |
+
"pool_aux": 0.01,
|
| 28 |
+
"pool_balance": 0.001,
|
| 29 |
+
"min_steps": 1,
|
| 30 |
+
"train_steps_mean": 8.0,
|
| 31 |
+
"train_depth": "halt:8/2/1",
|
| 32 |
+
"bptt_window": 24,
|
| 33 |
+
"ponder_beta": 0.01,
|
| 34 |
+
"halt_prior": 0.072,
|
| 35 |
+
"halt_thresh": 0.9,
|
| 36 |
+
"halt_freeze": true,
|
| 37 |
+
"adam_moments": "bf16",
|
| 38 |
+
"adam_beta2": 0.95,
|
| 39 |
+
"use_pool": true,
|
| 40 |
+
"pool_depth": 1,
|
| 41 |
+
"pool_max": 32
|
| 42 |
+
}
|
| 43 |
+
}
|
checkpoints/1388660k/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:07eed02ae17008e61757b22ff1c942ee5d2b403b3c27b3c351e9e8e83e9fbb7b
|
| 3 |
+
size 32638048
|