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

gguf-py : whitespace formatting fixes

Browse source
This commit is contained in:
Stanisław Szymczyk 2024-06-23 19:25:48 +02:00
parent a59f4f9eef
commit 843b1b7edc
1 changed files with 184 additions and 184 deletions
Download patch file Download diff file Expand all files Collapse all files

368
gguf-py/gguf/constants.py
View file

@ -82,35 +82,35 @@ class Keys:
TIME_STEP_RANK = "{arch}.ssm.time_step_rank" TIME_STEP_RANK = "{arch}.ssm.time_step_rank"
class Tokenizer: class Tokenizer:
MODEL = "tokenizer.ggml.model" MODEL = "tokenizer.ggml.model"
PRE = "tokenizer.ggml.pre" PRE = "tokenizer.ggml.pre"
LIST = "tokenizer.ggml.tokens" LIST = "tokenizer.ggml.tokens"
TOKEN_TYPE = "tokenizer.ggml.token_type" TOKEN_TYPE = "tokenizer.ggml.token_type"
TOKEN_TYPE_COUNT = "tokenizer.ggml.token_type_count" # for BERT-style token types TOKEN_TYPE_COUNT = "tokenizer.ggml.token_type_count" # for BERT-style token types
SCORES = "tokenizer.ggml.scores" SCORES = "tokenizer.ggml.scores"
MERGES = "tokenizer.ggml.merges" MERGES = "tokenizer.ggml.merges"
BOS_ID = "tokenizer.ggml.bos_token_id" BOS_ID = "tokenizer.ggml.bos_token_id"
EOS_ID = "tokenizer.ggml.eos_token_id" EOS_ID = "tokenizer.ggml.eos_token_id"
UNK_ID = "tokenizer.ggml.unknown_token_id" UNK_ID = "tokenizer.ggml.unknown_token_id"
SEP_ID = "tokenizer.ggml.seperator_token_id" SEP_ID = "tokenizer.ggml.seperator_token_id"
PAD_ID = "tokenizer.ggml.padding_token_id" PAD_ID = "tokenizer.ggml.padding_token_id"
CLS_ID = "tokenizer.ggml.cls_token_id" CLS_ID = "tokenizer.ggml.cls_token_id"
MASK_ID = "tokenizer.ggml.mask_token_id" MASK_ID = "tokenizer.ggml.mask_token_id"
ADD_BOS = "tokenizer.ggml.add_bos_token" ADD_BOS = "tokenizer.ggml.add_bos_token"
ADD_EOS = "tokenizer.ggml.add_eos_token" ADD_EOS = "tokenizer.ggml.add_eos_token"
ADD_PREFIX = "tokenizer.ggml.add_space_prefix" ADD_PREFIX = "tokenizer.ggml.add_space_prefix"
REMOVE_EXTRA_WS = "tokenizer.ggml.remove_extra_whitespaces" REMOVE_EXTRA_WS = "tokenizer.ggml.remove_extra_whitespaces"
PRECOMPILED_CHARSMAP = "tokenizer.ggml.precompiled_charsmap" PRECOMPILED_CHARSMAP = "tokenizer.ggml.precompiled_charsmap"
HF_JSON = "tokenizer.huggingface.json" HF_JSON = "tokenizer.huggingface.json"
RWKV = "tokenizer.rwkv.world" RWKV = "tokenizer.rwkv.world"
CHAT_TEMPLATE = "tokenizer.chat_template" CHAT_TEMPLATE = "tokenizer.chat_template"
CHAT_TEMPLATE_N = "tokenizer.chat_template.{name}" CHAT_TEMPLATE_N = "tokenizer.chat_template.{name}"
CHAT_TEMPLATES = "tokenizer.chat_templates" CHAT_TEMPLATES = "tokenizer.chat_templates"
# FIM/Infill special tokens constants # FIM/Infill special tokens constants
PREFIX_ID = "tokenizer.ggml.prefix_token_id" PREFIX_ID = "tokenizer.ggml.prefix_token_id"
SUFFIX_ID = "tokenizer.ggml.suffix_token_id" SUFFIX_ID = "tokenizer.ggml.suffix_token_id"
MIDDLE_ID = "tokenizer.ggml.middle_token_id" MIDDLE_ID = "tokenizer.ggml.middle_token_id"
EOT_ID = "tokenizer.ggml.eot_token_id" EOT_ID = "tokenizer.ggml.eot_token_id"
# #
@ -119,120 +119,120 @@ class Keys:
class MODEL_ARCH(IntEnum): class MODEL_ARCH(IntEnum):
LLAMA = auto() LLAMA = auto()
FALCON = auto() FALCON = auto()
BAICHUAN = auto() BAICHUAN = auto()
GROK = auto() GROK = auto()
GPT2 = auto() GPT2 = auto()
GPTJ = auto() GPTJ = auto()
GPTNEOX = auto() GPTNEOX = auto()
MPT = auto() MPT = auto()
STARCODER = auto() STARCODER = auto()
REFACT = auto() REFACT = auto()
BERT = auto() BERT = auto()
NOMIC_BERT = auto() NOMIC_BERT = auto()
JINA_BERT_V2 = auto() JINA_BERT_V2 = auto()
BLOOM = auto() BLOOM = auto()
STABLELM = auto() STABLELM = auto()
QWEN = auto() QWEN = auto()
QWEN2 = auto() QWEN2 = auto()
QWEN2MOE = auto() QWEN2MOE = auto()
PHI2 = auto() PHI2 = auto()
PHI3 = auto() PHI3 = auto()
PLAMO = auto() PLAMO = auto()
CODESHELL = auto() CODESHELL = auto()
ORION = auto() ORION = auto()
INTERNLM2 = auto() INTERNLM2 = auto()
MINICPM = auto() MINICPM = auto()
GEMMA = auto() GEMMA = auto()
STARCODER2 = auto() STARCODER2 = auto()
MAMBA = auto() MAMBA = auto()
XVERSE = auto() XVERSE = auto()
COMMAND_R = auto() COMMAND_R = auto()
DBRX = auto() DBRX = auto()
OLMO = auto() OLMO = auto()
ARCTIC = auto() ARCTIC = auto()
DEEPSEEK2 = auto() DEEPSEEK2 = auto()
T5 = auto() T5 = auto()
class MODEL_TENSOR(IntEnum): class MODEL_TENSOR(IntEnum):
TOKEN_EMBD = auto() TOKEN_EMBD = auto()
TOKEN_EMBD_NORM = auto() TOKEN_EMBD_NORM = auto()
TOKEN_TYPES = auto() TOKEN_TYPES = auto()
POS_EMBD = auto() POS_EMBD = auto()
OUTPUT = auto() OUTPUT = auto()
OUTPUT_NORM = auto() OUTPUT_NORM = auto()
ROPE_FREQS = auto() ROPE_FREQS = auto()
ROPE_FACTORS_LONG = auto() ROPE_FACTORS_LONG = auto()
ROPE_FACTORS_SHORT = auto() ROPE_FACTORS_SHORT = auto()
ATTN_Q = auto() ATTN_Q = auto()
ATTN_K = auto() ATTN_K = auto()
ATTN_V = auto() ATTN_V = auto()
ATTN_QKV = auto() ATTN_QKV = auto()
ATTN_OUT = auto() ATTN_OUT = auto()
ATTN_NORM = auto() ATTN_NORM = auto()
ATTN_NORM_2 = auto() ATTN_NORM_2 = auto()
ATTN_OUT_NORM = auto() ATTN_OUT_NORM = auto()
ATTN_ROT_EMBD = auto() ATTN_ROT_EMBD = auto()
FFN_GATE_INP = auto() FFN_GATE_INP = auto()
FFN_GATE_INP_SHEXP = auto() FFN_GATE_INP_SHEXP = auto()
FFN_NORM = auto() FFN_NORM = auto()
FFN_GATE = auto() FFN_GATE = auto()
FFN_DOWN = auto() FFN_DOWN = auto()
FFN_UP = auto() FFN_UP = auto()
FFN_ACT = auto() FFN_ACT = auto()
FFN_NORM_EXP = auto() FFN_NORM_EXP = auto()
FFN_GATE_EXP = auto() FFN_GATE_EXP = auto()
FFN_DOWN_EXP = auto() FFN_DOWN_EXP = auto()
FFN_UP_EXP = auto() FFN_UP_EXP = auto()
FFN_GATE_SHEXP = auto() FFN_GATE_SHEXP = auto()
FFN_DOWN_SHEXP = auto() FFN_DOWN_SHEXP = auto()
FFN_UP_SHEXP = auto() FFN_UP_SHEXP = auto()
ATTN_Q_NORM = auto() ATTN_Q_NORM = auto()
ATTN_K_NORM = auto() ATTN_K_NORM = auto()
LAYER_OUT_NORM = auto() LAYER_OUT_NORM = auto()
SSM_IN = auto() SSM_IN = auto()
SSM_CONV1D = auto() SSM_CONV1D = auto()
SSM_X = auto() SSM_X = auto()
SSM_DT = auto() SSM_DT = auto()
SSM_A = auto() SSM_A = auto()
SSM_D = auto() SSM_D = auto()
SSM_OUT = auto() SSM_OUT = auto()
ATTN_Q_A = auto() ATTN_Q_A = auto()
ATTN_Q_B = auto() ATTN_Q_B = auto()
ATTN_KV_A_MQA = auto() ATTN_KV_A_MQA = auto()
ATTN_KV_B = auto() ATTN_KV_B = auto()
ATTN_Q_A_NORM = auto() ATTN_Q_A_NORM = auto()
ATTN_KV_A_NORM = auto() ATTN_KV_A_NORM = auto()
DEC_ATTN_NORM = auto() DEC_ATTN_NORM = auto()
DEC_ATTN_Q = auto() DEC_ATTN_Q = auto()
DEC_ATTN_K = auto() DEC_ATTN_K = auto()
DEC_ATTN_V = auto() DEC_ATTN_V = auto()
DEC_ATTN_OUT = auto() DEC_ATTN_OUT = auto()
DEC_ATTN_REL_B = auto() DEC_ATTN_REL_B = auto()
DEC_CROSS_ATTN_NORM = auto() DEC_CROSS_ATTN_NORM = auto()
DEC_CROSS_ATTN_Q = auto() DEC_CROSS_ATTN_Q = auto()
DEC_CROSS_ATTN_K = auto() DEC_CROSS_ATTN_K = auto()
DEC_CROSS_ATTN_V = auto() DEC_CROSS_ATTN_V = auto()
DEC_CROSS_ATTN_OUT = auto() DEC_CROSS_ATTN_OUT = auto()
DEC_CROSS_ATTN_REL_B = auto() DEC_CROSS_ATTN_REL_B = auto()
DEC_FFN_NORM = auto() DEC_FFN_NORM = auto()
DEC_FFN_GATE = auto() DEC_FFN_GATE = auto()
DEC_FFN_DOWN = auto() DEC_FFN_DOWN = auto()
DEC_FFN_UP = auto() DEC_FFN_UP = auto()
DEC_OUTPUT_NORM = auto() DEC_OUTPUT_NORM = auto()
ENC_ATTN_NORM = auto() ENC_ATTN_NORM = auto()
ENC_ATTN_Q = auto() ENC_ATTN_Q = auto()
ENC_ATTN_K = auto() ENC_ATTN_K = auto()
ENC_ATTN_V = auto() ENC_ATTN_V = auto()
ENC_ATTN_OUT = auto() ENC_ATTN_OUT = auto()
ENC_ATTN_REL_B = auto() ENC_ATTN_REL_B = auto()
ENC_FFN_NORM = auto() ENC_FFN_NORM = auto()
ENC_FFN_GATE = auto() ENC_FFN_GATE = auto()
ENC_FFN_DOWN = auto() ENC_FFN_DOWN = auto()
ENC_FFN_UP = auto() ENC_FFN_UP = auto()
ENC_OUTPUT_NORM = auto() ENC_OUTPUT_NORM = auto()
MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = { MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
@ -274,54 +274,54 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
} }
TENSOR_NAMES: dict[MODEL_TENSOR, str] = { TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
MODEL_TENSOR.TOKEN_EMBD: "token_embd", MODEL_TENSOR.TOKEN_EMBD: "token_embd",
MODEL_TENSOR.TOKEN_EMBD_NORM: "token_embd_norm", MODEL_TENSOR.TOKEN_EMBD_NORM: "token_embd_norm",
MODEL_TENSOR.TOKEN_TYPES: "token_types", MODEL_TENSOR.TOKEN_TYPES: "token_types",
MODEL_TENSOR.POS_EMBD: "position_embd", MODEL_TENSOR.POS_EMBD: "position_embd",
MODEL_TENSOR.OUTPUT_NORM: "output_norm", MODEL_TENSOR.OUTPUT_NORM: "output_norm",
MODEL_TENSOR.OUTPUT: "output", MODEL_TENSOR.OUTPUT: "output",
MODEL_TENSOR.ROPE_FREQS: "rope_freqs", MODEL_TENSOR.ROPE_FREQS: "rope_freqs",
MODEL_TENSOR.ROPE_FACTORS_LONG: "rope_factors_long", MODEL_TENSOR.ROPE_FACTORS_LONG: "rope_factors_long",
MODEL_TENSOR.ROPE_FACTORS_SHORT: "rope_factors_short", MODEL_TENSOR.ROPE_FACTORS_SHORT: "rope_factors_short",
MODEL_TENSOR.ATTN_NORM: "blk.{bid}.attn_norm", MODEL_TENSOR.ATTN_NORM: "blk.{bid}.attn_norm",
MODEL_TENSOR.ATTN_NORM_2: "blk.{bid}.attn_norm_2", MODEL_TENSOR.ATTN_NORM_2: "blk.{bid}.attn_norm_2",
MODEL_TENSOR.ATTN_QKV: "blk.{bid}.attn_qkv", MODEL_TENSOR.ATTN_QKV: "blk.{bid}.attn_qkv",
MODEL_TENSOR.ATTN_Q: "blk.{bid}.attn_q", MODEL_TENSOR.ATTN_Q: "blk.{bid}.attn_q",
MODEL_TENSOR.ATTN_K: "blk.{bid}.attn_k", MODEL_TENSOR.ATTN_K: "blk.{bid}.attn_k",
MODEL_TENSOR.ATTN_V: "blk.{bid}.attn_v", MODEL_TENSOR.ATTN_V: "blk.{bid}.attn_v",
MODEL_TENSOR.ATTN_OUT: "blk.{bid}.attn_output", MODEL_TENSOR.ATTN_OUT: "blk.{bid}.attn_output",
MODEL_TENSOR.ATTN_ROT_EMBD: "blk.{bid}.attn_rot_embd", MODEL_TENSOR.ATTN_ROT_EMBD: "blk.{bid}.attn_rot_embd",
MODEL_TENSOR.ATTN_Q_NORM: "blk.{bid}.attn_q_norm", MODEL_TENSOR.ATTN_Q_NORM: "blk.{bid}.attn_q_norm",
MODEL_TENSOR.ATTN_K_NORM: "blk.{bid}.attn_k_norm", MODEL_TENSOR.ATTN_K_NORM: "blk.{bid}.attn_k_norm",
MODEL_TENSOR.ATTN_OUT_NORM: "blk.{bid}.attn_output_norm", MODEL_TENSOR.ATTN_OUT_NORM: "blk.{bid}.attn_output_norm",
MODEL_TENSOR.FFN_GATE_INP: "blk.{bid}.ffn_gate_inp", MODEL_TENSOR.FFN_GATE_INP: "blk.{bid}.ffn_gate_inp",
MODEL_TENSOR.FFN_GATE_INP_SHEXP: "blk.{bid}.ffn_gate_inp_shexp", MODEL_TENSOR.FFN_GATE_INP_SHEXP: "blk.{bid}.ffn_gate_inp_shexp",
MODEL_TENSOR.FFN_NORM: "blk.{bid}.ffn_norm", MODEL_TENSOR.FFN_NORM: "blk.{bid}.ffn_norm",
MODEL_TENSOR.FFN_GATE: "blk.{bid}.ffn_gate", MODEL_TENSOR.FFN_GATE: "blk.{bid}.ffn_gate",
MODEL_TENSOR.FFN_DOWN: "blk.{bid}.ffn_down", MODEL_TENSOR.FFN_DOWN: "blk.{bid}.ffn_down",
MODEL_TENSOR.FFN_UP: "blk.{bid}.ffn_up", MODEL_TENSOR.FFN_UP: "blk.{bid}.ffn_up",
MODEL_TENSOR.FFN_GATE_SHEXP: "blk.{bid}.ffn_gate_shexp", MODEL_TENSOR.FFN_GATE_SHEXP: "blk.{bid}.ffn_gate_shexp",
MODEL_TENSOR.FFN_DOWN_SHEXP: "blk.{bid}.ffn_down_shexp", MODEL_TENSOR.FFN_DOWN_SHEXP: "blk.{bid}.ffn_down_shexp",
MODEL_TENSOR.FFN_UP_SHEXP: "blk.{bid}.ffn_up_shexp", MODEL_TENSOR.FFN_UP_SHEXP: "blk.{bid}.ffn_up_shexp",
MODEL_TENSOR.FFN_ACT: "blk.{bid}.ffn", MODEL_TENSOR.FFN_ACT: "blk.{bid}.ffn",
MODEL_TENSOR.FFN_NORM_EXP: "blk.{bid}.ffn_norm_exps", MODEL_TENSOR.FFN_NORM_EXP: "blk.{bid}.ffn_norm_exps",
MODEL_TENSOR.FFN_GATE_EXP: "blk.{bid}.ffn_gate_exps", MODEL_TENSOR.FFN_GATE_EXP: "blk.{bid}.ffn_gate_exps",
MODEL_TENSOR.FFN_DOWN_EXP: "blk.{bid}.ffn_down_exps", MODEL_TENSOR.FFN_DOWN_EXP: "blk.{bid}.ffn_down_exps",
MODEL_TENSOR.FFN_UP_EXP: "blk.{bid}.ffn_up_exps", MODEL_TENSOR.FFN_UP_EXP: "blk.{bid}.ffn_up_exps",
MODEL_TENSOR.LAYER_OUT_NORM: "blk.{bid}.layer_output_norm", MODEL_TENSOR.LAYER_OUT_NORM: "blk.{bid}.layer_output_norm",
MODEL_TENSOR.SSM_IN: "blk.{bid}.ssm_in", MODEL_TENSOR.SSM_IN: "blk.{bid}.ssm_in",
MODEL_TENSOR.SSM_CONV1D: "blk.{bid}.ssm_conv1d", MODEL_TENSOR.SSM_CONV1D: "blk.{bid}.ssm_conv1d",
MODEL_TENSOR.SSM_X: "blk.{bid}.ssm_x", MODEL_TENSOR.SSM_X: "blk.{bid}.ssm_x",
MODEL_TENSOR.SSM_DT: "blk.{bid}.ssm_dt", MODEL_TENSOR.SSM_DT: "blk.{bid}.ssm_dt",
MODEL_TENSOR.SSM_A: "blk.{bid}.ssm_a", MODEL_TENSOR.SSM_A: "blk.{bid}.ssm_a",
MODEL_TENSOR.SSM_D: "blk.{bid}.ssm_d", MODEL_TENSOR.SSM_D: "blk.{bid}.ssm_d",
MODEL_TENSOR.SSM_OUT: "blk.{bid}.ssm_out", MODEL_TENSOR.SSM_OUT: "blk.{bid}.ssm_out",
MODEL_TENSOR.ATTN_Q_A: "blk.{bid}.attn_q_a", MODEL_TENSOR.ATTN_Q_A: "blk.{bid}.attn_q_a",
MODEL_TENSOR.ATTN_Q_B: "blk.{bid}.attn_q_b", MODEL_TENSOR.ATTN_Q_B: "blk.{bid}.attn_q_b",
MODEL_TENSOR.ATTN_KV_A_MQA: "blk.{bid}.attn_kv_a_mqa", MODEL_TENSOR.ATTN_KV_A_MQA: "blk.{bid}.attn_kv_a_mqa",
MODEL_TENSOR.ATTN_KV_B: "blk.{bid}.attn_kv_b", MODEL_TENSOR.ATTN_KV_B: "blk.{bid}.attn_kv_b",
MODEL_TENSOR.ATTN_Q_A_NORM: "blk.{bid}.attn_q_a_norm", MODEL_TENSOR.ATTN_Q_A_NORM: "blk.{bid}.attn_q_a_norm",
MODEL_TENSOR.ATTN_KV_A_NORM: "blk.{bid}.attn_kv_a_norm", MODEL_TENSOR.ATTN_KV_A_NORM: "blk.{bid}.attn_kv_a_norm",
MODEL_TENSOR.DEC_ATTN_NORM: "dec.blk.{bid}.attn_norm", MODEL_TENSOR.DEC_ATTN_NORM: "dec.blk.{bid}.attn_norm",
MODEL_TENSOR.DEC_ATTN_Q: "dec.blk.{bid}.attn_q", MODEL_TENSOR.DEC_ATTN_Q: "dec.blk.{bid}.attn_q",
MODEL_TENSOR.DEC_ATTN_K: "dec.blk.{bid}.attn_k", MODEL_TENSOR.DEC_ATTN_K: "dec.blk.{bid}.attn_k",