Upload folder using huggingface_hub
Browse files- final_model.pt +3 -0
- model_config.json +1 -0
final_model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ddbb3b999bfba192b61a66c0bfca2992ad475fccbbcd2b8c8ed05b217975c38e
|
| 3 |
+
size 3372996880
|
model_config.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"block_size": 1024, "vocab_size": 4096, "n_layer": 4, "n_head": 256, "d_head": 16, "d_model": 4096, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v,attn_k,attn_q", "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 16384, "enable_bigram_table": false, "learnable_bigram_table": false, "d_pos_emb": null, "dropout_cat_pos_emb": false, "sinusoidal_cat_pos_emb": false, "enable_sparse_kernels": false, "flash": true, "sink": true}
|