vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Fix whitespaces

Browse source
This commit is contained in:
Julius Arkenberg 2024-03-21 19:59:15 +00:00
parent 6052e3b3a7
commit 81ce9df3ee
3 changed files with 38 additions and 44 deletions
Download patch file Download diff file Expand all files Collapse all files

2
convert-hf-to-gguf.py
View file

@ -1077,8 +1077,6 @@ class GrokModel(Model):
self.gguf_writer.add_name("Grok") self.gguf_writer.add_name("Grok")
@Model.register("MiniCPMForCausalLM") @Model.register("MiniCPMForCausalLM")
class MiniCPMModel(Model): class MiniCPMModel(Model):
model_arch = gguf.MODEL_ARCH.MINICPM model_arch = gguf.MODEL_ARCH.MINICPM

4
gguf-py/gguf/tensor_mapping.py
View file

@ -168,7 +168,7 @@ class TensorNameMap:
"model.layers.layers.{bid}.self_attn.o_proj", # plamo "model.layers.layers.{bid}.self_attn.o_proj", # plamo
"model.layers.{bid}.attention.wo", # internlm2 "model.layers.{bid}.attention.wo", # internlm2
"encoder.layers.{bid}.attn.out_proj", # nomic-bert "encoder.layers.{bid}.attn.out_proj", # nomic-bert
"transformer.decoder_layer.{bid}.multi_head_attention.linear" # Grok "transformer.decoder_layer.{bid}.multi_head_attention.linear"# Grok
), ),
# Attention output norm # Attention output norm
@ -198,9 +198,7 @@ class TensorNameMap:
"model.layers.{bid}.ln2", # yi "model.layers.{bid}.ln2", # yi
"h.{bid}.ln_2", # gpt2 "h.{bid}.ln_2", # gpt2
"model.layers.{bid}.ffn_norm", # internlm2 "model.layers.{bid}.ffn_norm", # internlm2
"transformer.decoder_layer.{bid}.rms_norm_2", # Grok "transformer.decoder_layer.{bid}.rms_norm_2", # Grok
), ),
MODEL_TENSOR.FFN_GATE_INP: ( MODEL_TENSOR.FFN_GATE_INP: (

2
llama.cpp
View file

@ -4346,8 +4346,6 @@ static bool llm_load_tensors(
layer.ffn_up_exp[x] = ml.create_tensor(ctx_split, tn(LLM_TENSOR_FFN_UP_EXP, "weight", i, x), {n_embd, n_ff}); layer.ffn_up_exp[x] = ml.create_tensor(ctx_split, tn(LLM_TENSOR_FFN_UP_EXP, "weight", i, x), {n_embd, n_ff});
} }
layer.layer_out_norm = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_LAYER_OUT_NORM, "weight", i), {n_embd}); layer.layer_out_norm = ml.create_tensor(ctx_layer, tn(LLM_TENSOR_LAYER_OUT_NORM, "weight", i), {n_embd});
} }
} break; } break;