Remove old-layout metadata orphaned by MODEL_SPEC.md restructure
Browse files
config.json
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"model_type": "smollm2"
|
| 3 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-1.7B/original/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 2048,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 8192,
|
| 5 |
-
"n_heads": 32,
|
| 6 |
-
"n_kv_heads": 32,
|
| 7 |
-
"n_layers": 24,
|
| 8 |
-
"norm_eps": 1e-5,
|
| 9 |
-
"rope_theta": 130000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-1.7B/quantized/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 2048,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 8192,
|
| 5 |
-
"n_heads": 32,
|
| 6 |
-
"n_kv_heads": 32,
|
| 7 |
-
"n_layers": 24,
|
| 8 |
-
"norm_eps": 1e-5,
|
| 9 |
-
"rope_theta": 130000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-135M/original/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 576,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 1536,
|
| 5 |
-
"n_heads": 9,
|
| 6 |
-
"n_kv_heads": 3,
|
| 7 |
-
"n_layers": 30,
|
| 8 |
-
"norm_eps": 1e-05,
|
| 9 |
-
"rope_theta": 10000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-135M/quantized/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 576,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 1536,
|
| 5 |
-
"n_heads": 9,
|
| 6 |
-
"n_kv_heads": 3,
|
| 7 |
-
"n_layers": 30,
|
| 8 |
-
"norm_eps": 1e-05,
|
| 9 |
-
"rope_theta": 10000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-360M/original/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 960,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 2560,
|
| 5 |
-
"n_heads": 15,
|
| 6 |
-
"n_kv_heads": 5,
|
| 7 |
-
"n_layers": 32,
|
| 8 |
-
"norm_eps": 1e-5,
|
| 9 |
-
"rope_theta": 100000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
smolLm-2-360M/quantized/config.json
DELETED
|
@@ -1,14 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"dim": 960,
|
| 3 |
-
"ffn_dim_multiplier": 1,
|
| 4 |
-
"hidden_dim": 2560,
|
| 5 |
-
"n_heads": 15,
|
| 6 |
-
"n_kv_heads": 5,
|
| 7 |
-
"n_layers": 32,
|
| 8 |
-
"norm_eps": 1e-5,
|
| 9 |
-
"rope_theta": 100000.0,
|
| 10 |
-
"use_scaled_rope": false,
|
| 11 |
-
"vocab_size": 49152,
|
| 12 |
-
"use_hf_rope": false,
|
| 13 |
-
"attention_qkv_bias": false
|
| 14 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|