Merge branch 'master' into compilade/refactor-kv-cache
This commit is contained in:
Francis Couture-Harpin
commit
bc320ef66d
395 changed files with 57725 additions and 169970 deletions
|
|
@ -10,10 +10,10 @@ class TensorNameMap:
|
|||
# Token embeddings
|
||||
MODEL_TENSOR.TOKEN_EMBD: (
|
||||
"gpt_neox.embed_in", # gptneox
|
||||
"transformer.wte", # gpt2 gpt-j mpt refact qwen dbrx jais
|
||||
"transformer.wte", # gpt2 gpt-j mpt refact qwen dbrx jais exaone
|
||||
"transformer.word_embeddings", # falcon
|
||||
"word_embeddings", # bloom
|
||||
"model.embed_tokens", # llama-hf
|
||||
"model.embed_tokens", # llama-hf nemotron
|
||||
"tok_embeddings", # llama-pth
|
||||
"embeddings.word_embeddings", # bert nomic-bert
|
||||
"language_model.embedding.word_embeddings", # persimmon
|
||||
|
|
@ -24,6 +24,7 @@ class TensorNameMap:
|
|||
"backbone.embedding", # mamba
|
||||
"backbone.embeddings", # mamba-hf
|
||||
"transformer.in_out_embed", # Grok
|
||||
"embedding.word_embeddings", # chatglm
|
||||
"transformer.token_embeddings", # openelm
|
||||
"shared", # t5
|
||||
),
|
||||
|
|
@ -51,16 +52,17 @@ class TensorNameMap:
|
|||
# Output
|
||||
MODEL_TENSOR.OUTPUT: (
|
||||
"embed_out", # gptneox
|
||||
"lm_head", # gpt2 mpt falcon llama-hf baichuan qwen mamba dbrx jais
|
||||
"lm_head", # gpt2 mpt falcon llama-hf baichuan qwen mamba dbrx jais nemotron exaone
|
||||
"output", # llama-pth bloom internlm2
|
||||
"word_embeddings_for_head", # persimmon
|
||||
"lm_head.linear", # phi2
|
||||
"output_layer", # chatglm
|
||||
),
|
||||
|
||||
# Output norm
|
||||
MODEL_TENSOR.OUTPUT_NORM: (
|
||||
"gpt_neox.final_layer_norm", # gptneox
|
||||
"transformer.ln_f", # gpt2 gpt-j falcon jais
|
||||
"transformer.ln_f", # gpt2 gpt-j falcon jais exaone
|
||||
"model.norm", # llama-hf baichuan internlm2
|
||||
"norm", # llama-pth
|
||||
"transformer.norm_f", # mpt dbrx
|
||||
|
|
@ -71,12 +73,15 @@ class TensorNameMap:
|
|||
"model.norm_f", # mamba-qbert
|
||||
"backbone.norm_f", # mamba
|
||||
"transformer.rms_norm", # Grok
|
||||
"encoder.final_layernorm", # chatglm
|
||||
"transformer.norm", # openelm
|
||||
"model.norm", # nemotron
|
||||
),
|
||||
|
||||
# Rope frequencies
|
||||
MODEL_TENSOR.ROPE_FREQS: (
|
||||
"rope.freqs", # llama-pth
|
||||
"rotary_pos_emb.inv_freq", # chatglm
|
||||
),
|
||||
}
|
||||
|
||||
|
|
@ -84,12 +89,12 @@ class TensorNameMap:
|
|||
# Attention norm
|
||||
MODEL_TENSOR.ATTN_NORM: (
|
||||
"gpt_neox.layers.{bid}.input_layernorm", # gptneox
|
||||
"transformer.h.{bid}.ln_1", # gpt2 gpt-j refact qwen jais
|
||||
"transformer.h.{bid}.ln_1", # gpt2 gpt-j refact qwen jais exaone
|
||||
"transformer.blocks.{bid}.norm_1", # mpt
|
||||
"transformer.h.{bid}.input_layernorm", # falcon7b
|
||||
"h.{bid}.input_layernorm", # bloom
|
||||
"transformer.h.{bid}.ln_mlp", # falcon40b
|
||||
"model.layers.{bid}.input_layernorm", # llama-hf
|
||||
"model.layers.{bid}.input_layernorm", # llama-hf nemotron
|
||||
"layers.{bid}.attention_norm", # llama-pth
|
||||
"language_model.encoder.layers.{bid}.input_layernorm", # persimmon
|
||||
"model.layers.{bid}.ln1", # yi
|
||||
|
|
@ -101,6 +106,7 @@ class TensorNameMap:
|
|||
"backbone.layers.{bid}.norm", # mamba
|
||||
"transformer.decoder_layer.{bid}.rms_norm", # Grok
|
||||
"transformer.blocks.{bid}.norm_attn_norm.norm_1", # dbrx
|
||||
"encoder.layers.{bid}.input_layernorm", # chatglm
|
||||
"transformer.layers.{bid}.attn_norm", # openelm
|
||||
),
|
||||
|
||||
|
|
@ -124,42 +130,46 @@ class TensorNameMap:
|
|||
"transformer.h.{bid}.mixer.Wqkv", # phi2
|
||||
"encoder.layers.{bid}.attn.Wqkv", # nomic-bert
|
||||
"model.layers.{bid}.self_attn.qkv_proj", # phi3
|
||||
"encoder.layers.{bid}.self_attention.query_key_value", # chatglm
|
||||
"transformer.layers.{bid}.attn.qkv_proj", # openelm
|
||||
),
|
||||
|
||||
# Attention query
|
||||
MODEL_TENSOR.ATTN_Q: (
|
||||
"model.layers.{bid}.self_attn.q_proj", # llama-hf
|
||||
"model.layers.{bid}.self_attn.q_proj", # llama-hf nemotron
|
||||
"layers.{bid}.attention.wq", # llama-pth
|
||||
"encoder.layer.{bid}.attention.self.query", # bert
|
||||
"transformer.h.{bid}.attn.q_proj", # gpt-j
|
||||
"model.layers.layers.{bid}.self_attn.q_proj", # plamo
|
||||
"model.layers.{bid}.attention.wq", # internlm2
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.query" # Grok
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.query",# Grok
|
||||
"transformer.h.{bid}.attn.attention.q_proj", # exaone
|
||||
),
|
||||
|
||||
# Attention key
|
||||
MODEL_TENSOR.ATTN_K: (
|
||||
"model.layers.{bid}.self_attn.k_proj", # llama-hf
|
||||
"model.layers.{bid}.self_attn.k_proj", # llama-hf nemotron
|
||||
"layers.{bid}.attention.wk", # llama-pth
|
||||
"encoder.layer.{bid}.attention.self.key", # bert
|
||||
"transformer.h.{bid}.attn.k_proj", # gpt-j
|
||||
"transformer.h.{bid}.attn.k", # refact
|
||||
"model.layers.layers.{bid}.self_attn.k_proj", # plamo
|
||||
"model.layers.{bid}.attention.wk", # internlm2
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.key" # Grok
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.key",# Grok
|
||||
"transformer.h.{bid}.attn.attention.k_proj", # exaone
|
||||
),
|
||||
|
||||
# Attention value
|
||||
MODEL_TENSOR.ATTN_V: (
|
||||
"model.layers.{bid}.self_attn.v_proj", # llama-hf
|
||||
"model.layers.{bid}.self_attn.v_proj", # llama-hf nemotron
|
||||
"layers.{bid}.attention.wv", # llama-pth
|
||||
"encoder.layer.{bid}.attention.self.value", # bert
|
||||
"transformer.h.{bid}.attn.v_proj", # gpt-j
|
||||
"transformer.h.{bid}.attn.v", # refact
|
||||
"model.layers.layers.{bid}.self_attn.v_proj", # plamo
|
||||
"model.layers.{bid}.attention.wv", # internlm2
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.value" # Grok
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.value",# Grok
|
||||
"transformer.h.{bid}.attn.attention.v_proj", # exaone
|
||||
),
|
||||
|
||||
# Attention output
|
||||
|
|
@ -169,7 +179,7 @@ class TensorNameMap:
|
|||
"transformer.blocks.{bid}.attn.out_proj", # mpt
|
||||
"transformer.h.{bid}.self_attention.dense", # falcon
|
||||
"h.{bid}.self_attention.dense", # bloom
|
||||
"model.layers.{bid}.self_attn.o_proj", # llama-hf
|
||||
"model.layers.{bid}.self_attn.o_proj", # llama-hf nemotron
|
||||
"layers.{bid}.attention.wo", # llama-pth
|
||||
"encoder.layer.{bid}.attention.output.dense", # bert
|
||||
"transformer.h.{bid}.attn.out_proj", # gpt-j
|
||||
|
|
@ -182,7 +192,9 @@ class TensorNameMap:
|
|||
"encoder.layers.{bid}.attn.out_proj", # nomic-bert
|
||||
"transformer.decoder_layer.{bid}.multi_head_attention.linear", # Grok
|
||||
"transformer.blocks.{bid}.norm_attn_norm.attn.out_proj", # dbrx
|
||||
"encoder.layers.{bid}.self_attention.dense", # chatglm
|
||||
"transformer.layers.{bid}.attn.out_proj", # openelm
|
||||
"transformer.h.{bid}.attn.attention.out_proj", # exaone
|
||||
),
|
||||
|
||||
# Attention output norm
|
||||
|
|
@ -208,16 +220,17 @@ class TensorNameMap:
|
|||
# Feed-forward norm
|
||||
MODEL_TENSOR.FFN_NORM: (
|
||||
"gpt_neox.layers.{bid}.post_attention_layernorm", # gptneox
|
||||
"transformer.h.{bid}.ln_2", # gpt2 refact qwen jais
|
||||
"transformer.h.{bid}.ln_2", # gpt2 refact qwen jais exaone
|
||||
"h.{bid}.post_attention_layernorm", # bloom
|
||||
"transformer.blocks.{bid}.norm_2", # mpt
|
||||
"model.layers.{bid}.post_attention_layernorm", # llama-hf
|
||||
"model.layers.{bid}.post_attention_layernorm", # llama-hf nemotron
|
||||
"layers.{bid}.ffn_norm", # llama-pth
|
||||
"language_model.encoder.layers.{bid}.post_attention_layernorm", # persimmon
|
||||
"model.layers.{bid}.ln2", # yi
|
||||
"h.{bid}.ln_2", # gpt2
|
||||
"model.layers.{bid}.ffn_norm", # internlm2
|
||||
"transformer.decoder_layer.{bid}.rms_norm_2", # Grok
|
||||
"encoder.layers.{bid}.post_attention_layernorm", # chatglm
|
||||
"transformer.layers.{bid}.ffn_norm", # openelm
|
||||
"model.layers.{bid}.pre_ff_layernorm", # jamba
|
||||
"model.layers.{bid}.pre_moe_layernorm", # mini-jamba
|
||||
|
|
@ -253,7 +266,7 @@ class TensorNameMap:
|
|||
"transformer.blocks.{bid}.ffn.up_proj", # mpt
|
||||
"transformer.h.{bid}.mlp.dense_h_to_4h", # falcon
|
||||
"h.{bid}.mlp.dense_h_to_4h", # bloom
|
||||
"model.layers.{bid}.mlp.up_proj", # llama-hf refact
|
||||
"model.layers.{bid}.mlp.up_proj", # llama-hf refact nemotron
|
||||
"layers.{bid}.feed_forward.w3", # llama-pth
|
||||
"encoder.layer.{bid}.intermediate.dense", # bert
|
||||
"transformer.h.{bid}.mlp.fc_in", # gpt-j
|
||||
|
|
@ -272,6 +285,8 @@ class TensorNameMap:
|
|||
"encoder.layer.{bid}.mlp.gated_layers_v", # jina-bert-v2
|
||||
"model.layers.{bid}.residual_mlp.w3", # arctic
|
||||
"model.layers.{bid}.feed_forward.up_proj", # jamba
|
||||
"encoder.layers.{bid}.mlp.dense_h_to_4h", # chatglm
|
||||
"transformer.h.{bid}.mlp.c_fc_1", # exaone
|
||||
),
|
||||
|
||||
MODEL_TENSOR.FFN_UP_EXP: (
|
||||
|
|
@ -304,6 +319,7 @@ class TensorNameMap:
|
|||
"transformer.h.{bid}.mlp.linear_1", # refact
|
||||
"model.layers.{bid}.residual_mlp.w1", # arctic
|
||||
"model.layers.{bid}.feed_forward.gate_proj", # jamba
|
||||
"transformer.h.{bid}.mlp.c_fc_0", # exaone
|
||||
),
|
||||
|
||||
MODEL_TENSOR.FFN_GATE_EXP: (
|
||||
|
|
@ -325,7 +341,7 @@ class TensorNameMap:
|
|||
"transformer.blocks.{bid}.ffn.down_proj", # mpt
|
||||
"transformer.h.{bid}.mlp.dense_4h_to_h", # falcon
|
||||
"h.{bid}.mlp.dense_4h_to_h", # bloom
|
||||
"model.layers.{bid}.mlp.down_proj", # llama-hf
|
||||
"model.layers.{bid}.mlp.down_proj", # llama-hf nemotron
|
||||
"layers.{bid}.feed_forward.w2", # llama-pth
|
||||
"encoder.layer.{bid}.output.dense", # bert
|
||||
"transformer.h.{bid}.mlp.fc_out", # gpt-j
|
||||
|
|
@ -343,6 +359,8 @@ class TensorNameMap:
|
|||
"model.layers.{bid}.residual_mlp.w2", # arctic
|
||||
"encoder.layer.{bid}.mlp.down_layer", # jina-bert-v2
|
||||
"model.layers.{bid}.feed_forward.down_proj", # jamba
|
||||
"encoder.layers.{bid}.mlp.dense_4h_to_h", # chatglm
|
||||
"model.layers.h.{bid}.mlp.c_proj", # exaone
|
||||
),
|
||||
|
||||
MODEL_TENSOR.FFN_DOWN_EXP: (
|
||||
|
|
@ -619,14 +637,12 @@ class TensorNameMap:
|
|||
for tensor, keys in self.block_mappings_cfg.items():
|
||||
if tensor not in MODEL_TENSORS[arch]:
|
||||
continue
|
||||
# TODO: make this configurable
|
||||
n_experts = 160
|
||||
for xid in range(n_experts):
|
||||
tensor_name = TENSOR_NAMES[tensor].format(bid = bid, xid = xid)
|
||||
self.mapping[tensor_name] = (tensor, tensor_name)
|
||||
for key in keys:
|
||||
key = key.format(bid = bid, xid = xid)
|
||||
self.mapping[key] = (tensor, tensor_name)
|
||||
|
||||
tensor_name = TENSOR_NAMES[tensor].format(bid = bid)
|
||||
self.mapping[tensor_name] = (tensor, tensor_name)
|
||||
for key in keys:
|
||||
key = key.format(bid = bid)
|
||||
self.mapping[key] = (tensor, tensor_name)
|
||||
|
||||
def get_type_and_name(self, key: str, try_suffixes: Sequence[str] = ()) -> tuple[MODEL_TENSOR, str] | None:
|
||||
result = self.mapping.get(key)
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue