Update modeling_neollm.py
Browse files- modeling_neollm.py +1 -1
modeling_neollm.py
CHANGED
|
@@ -1200,6 +1200,7 @@ class NeoLLMForCausalLM(NeoLLMPreTrainedModel, GenerationMixin):
|
|
| 1200 |
Note on LM head: Following "Learnable Multipliers" paper recommendations,
|
| 1201 |
the output projection (lm_head) does NOT include learnable multipliers.
|
| 1202 |
"""
|
|
|
|
| 1203 |
|
| 1204 |
def __init__(self, config):
|
| 1205 |
super().__init__(config)
|
|
@@ -1210,7 +1211,6 @@ class NeoLLMForCausalLM(NeoLLMPreTrainedModel, GenerationMixin):
|
|
| 1210 |
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
|
| 1211 |
|
| 1212 |
self.post_init()
|
| 1213 |
-
_tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
|
| 1214 |
|
| 1215 |
def forward(
|
| 1216 |
self,
|
|
|
|
| 1200 |
Note on LM head: Following "Learnable Multipliers" paper recommendations,
|
| 1201 |
the output projection (lm_head) does NOT include learnable multipliers.
|
| 1202 |
"""
|
| 1203 |
+
_tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
|
| 1204 |
|
| 1205 |
def __init__(self, config):
|
| 1206 |
super().__init__(config)
|
|
|
|
| 1211 |
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
|
| 1212 |
|
| 1213 |
self.post_init()
|
|
|
|
| 1214 |
|
| 1215 |
def forward(
|
| 1216 |
self,
|