CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-10 10:27:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into compilade/imatrix-batched-chunks

Browse Source
This commit is contained in:
Francis Couture-Harpin
2025-06-18 16:32:35 -04:00
parent 1d19025909 8d94713654
commit 2c0945027a
687 changed files with 177746 additions and 45543 deletions
Download Patch File Download Diff File Expand all files Collapse all files

344
gguf-py/gguf/constants.py
View File

@@ -104,6 +104,7 @@ class Keys:
EXPERT_WEIGHTS_SCALE = "{arch}.expert_weights_scale"
EXPERT_WEIGHTS_NORM = "{arch}.expert_weights_norm"
EXPERT_GATING_FUNC = "{arch}.expert_gating_func"
MOE_EVERY_N_LAYERS = "{arch}.moe_every_n_layers"
POOLING_TYPE = "{arch}.pooling_type"
LOGIT_SCALE = "{arch}.logit_scale"
DECODER_START_TOKEN_ID = "{arch}.decoder_start_token_id"
@@ -139,6 +140,8 @@ class Keys:
REL_BUCKETS_COUNT = "{arch}.attention.relative_buckets_count"
SLIDING_WINDOW = "{arch}.attention.sliding_window"
SCALE = "{arch}.attention.scale"
KEY_LENGTH_MLA = "{arch}.attention.key_length_mla"
VALUE_LENGTH_MLA = "{arch}.attention.value_length_mla"
class Rope:
DIMENSION_COUNT = "{arch}.rope.dimension_count"
@@ -174,6 +177,9 @@ class Keys:
EMBEDDING_LENGTH = "{arch}.convnext.embedding_length"
BLOCK_COUNT = "{arch}.convnext.block_count"
class Classifier:
OUTPUT_LABELS = "{arch}.classifier.output_labels"
class Tokenizer:
MODEL = "tokenizer.ggml.model"
PRE = "tokenizer.ggml.pre"
@@ -221,6 +227,46 @@ class Keys:
CHUNK_SIZE = "imatrix.chunk_size"
DATASETS = "imatrix.datasets"
class Clip:
PROJECTOR_TYPE = "clip.projector_type"
HAS_VISION_ENCODER = "clip.has_vision_encoder"
HAS_AUDIO_ENCODER = "clip.has_audio_encoder"
HAS_LLAVA_PROJECTOR = "clip.has_llava_projector"
class ClipVision:
IMAGE_SIZE = "clip.vision.image_size"
PATCH_SIZE = "clip.vision.patch_size"
EMBEDDING_LENGTH = "clip.vision.embedding_length"
FEED_FORWARD_LENGTH = "clip.vision.feed_forward_length"
PROJECTION_DIM = "clip.vision.projection_dim"
BLOCK_COUNT = "clip.vision.block_count"
IMAGE_MEAN = "clip.vision.image_mean"
IMAGE_STD = "clip.vision.image_std"
SPATIAL_MERGE_SIZE = "clip.vision.spatial_merge_size"
USE_GELU = "clip.use_gelu"
USE_SILU = "clip.use_silu"
N_WA_PATTERN = "clip.vision.n_wa_pattern" # used by qwen2.5vl
class Attention:
HEAD_COUNT = "clip.vision.attention.head_count"
LAYERNORM_EPS = "clip.vision.attention.layer_norm_epsilon"
class Projector:
SCALE_FACTOR = "clip.vision.projector.scale_factor"
class ClipAudio:
NUM_MEL_BINS = "clip.audio.num_mel_bins"
EMBEDDING_LENGTH = "clip.audio.embedding_length"
FEED_FORWARD_LENGTH = "clip.audio.feed_forward_length"
PROJECTION_DIM = "clip.audio.projection_dim"
BLOCK_COUNT = "clip.audio.block_count"
class Attention:
HEAD_COUNT = "clip.audio.attention.head_count"
LAYERNORM_EPS = "clip.audio.attention.layer_norm_epsilon"
class Projector:
STACK_FACTOR = "clip.audio.projector.stack_factor"
#
# recommended mapping of model tensor names for storage in gguf
@@ -231,9 +277,11 @@ class GGUFType:
MODEL = "model"
ADAPTER = "adapter"
IMATRIX = "imatrix"
MMPROJ = "mmproj" # dummy, unused for now
class MODEL_ARCH(IntEnum):
MMPROJ = auto() # dummy arch for clip.cpp
LLAMA = auto()
LLAMA4 = auto()
DECI = auto()
@@ -248,6 +296,8 @@ class MODEL_ARCH(IntEnum):
REFACT = auto()
BERT = auto()
NOMIC_BERT = auto()
NOMIC_BERT_MOE = auto()
NEO_BERT = auto()
JINA_BERT_V2 = auto()
BLOOM = auto()
STABLELM = auto()
@@ -300,6 +350,18 @@ class MODEL_ARCH(IntEnum):
WAVTOKENIZER_DEC = auto()
PLM = auto()
BAILINGMOE = auto()
DOTS1 = auto()
ARCEE = auto()
class VISION_PROJECTOR_TYPE(IntEnum):
MLP = auto()
LDP = auto()
LDPV2 = auto()
RESAMPLER = auto()
GLM_EDGE = auto()
MERGER = auto()
GEMMA3 = auto()
class MODEL_TENSOR(IntEnum):
@@ -389,6 +451,8 @@ class MODEL_TENSOR(IntEnum):
ATTN_Q_B = auto()
ATTN_KV_A_MQA = auto()
ATTN_KV_B = auto()
ATTN_K_B = auto()
ATTN_V_B = auto()
ATTN_Q_A_NORM = auto()
ATTN_KV_A_NORM = auto()
FFN_SUB_NORM = auto()
@@ -439,9 +503,68 @@ class MODEL_TENSOR(IntEnum):
POSNET_ATTN_K = auto()
POSNET_ATTN_V = auto()
POSNET_ATTN_OUT = auto()
# vision
V_MMPROJ = auto()
V_MMPROJ_FC = auto()
V_MMPROJ_MLP = auto()
V_MMPROJ_PEG = auto()
V_ENC_EMBD_CLS = auto()
V_ENC_EMBD_PATCH = auto()
V_ENC_EMBD_POS = auto()
V_ENC_INPUT_NORM = auto()
V_ENC_ATTN_Q = auto()
V_ENC_ATTN_Q_NORM = auto()
V_ENC_ATTN_K = auto()
V_ENC_ATTN_K_NORM = auto()
V_ENC_ATTN_V = auto()
V_ENC_ATTN_O = auto()
V_ENC_ATTN_O_NORM = auto()
V_ENC_POST_ATTN_NORM = auto()
V_ENC_FFN_UP = auto()
V_ENC_FFN_GATE = auto()
V_ENC_FFN_DOWN = auto()
V_LAYER_SCALE_1 = auto()
V_LAYER_SCALE_2 = auto()
V_PRE_NORM = auto()
V_POST_NORM = auto()
V_MM_INP_NORM = auto()
V_MM_INP_PROJ = auto() # gemma3
V_MM_SOFT_EMB_NORM = auto() # gemma3
V_RESMPL_POS_EMBD_K = auto() # minicpmv
V_RESMPL_ATTN_Q = auto() # minicpmv
V_RESMPL_ATTN_K = auto() # minicpmv
V_RESMPL_ATTN_V = auto() # minicpmv
V_RESMPL_ATTN_OUT = auto() # minicpmv
V_RESMPL_KV = auto() # minicpmv
V_RESMPL_KV_NORM = auto() # minicpmv
V_RESMPL_POST_NORM = auto() # minicpmv
V_RESMPL_Q_NORM = auto() # minicpmv
V_RESMPL_PROJ = auto() # minicpmv
V_RESMPL_QUERY = auto() # minicpmv
V_TOK_EMBD_IMG_BREAK = auto() # pixtral
V_MM_PATCH_MERGER = auto() # mistral small 3.1
# audio (mtmd)
A_ENC_EMBD_POS = auto()
A_ENC_CONV1D = auto()
A_PRE_NORM = auto()
A_POST_NORM = auto()
A_ENC_ATTN_Q = auto()
A_ENC_ATTN_K = auto()
A_ENC_ATTN_V = auto()
A_ENC_INPUT_NORM = auto()
A_ENC_OUTPUT = auto()
A_ENC_OUTPUT_NORM = auto()
A_ENC_FFN_UP = auto()
A_ENC_FFN_GATE = auto()
A_ENC_FFN_DOWN = auto()
A_MMPROJ = auto()
A_MMPROJ_FC = auto()
A_MM_NORM_PRE = auto()
A_MM_NORM_MID = auto()
MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.MMPROJ: "clip", # dummy arch for clip.cpp
MODEL_ARCH.LLAMA: "llama",
MODEL_ARCH.LLAMA4: "llama4",
MODEL_ARCH.DECI: "deci",
@@ -456,6 +579,8 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.REFACT: "refact",
MODEL_ARCH.BERT: "bert",
MODEL_ARCH.NOMIC_BERT: "nomic-bert",
MODEL_ARCH.NOMIC_BERT_MOE: "nomic-bert-moe",
MODEL_ARCH.NEO_BERT: "neo-bert",
MODEL_ARCH.JINA_BERT_V2: "jina-bert-v2",
MODEL_ARCH.BLOOM: "bloom",
MODEL_ARCH.STABLELM: "stablelm",
@@ -508,6 +633,18 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = {
MODEL_ARCH.WAVTOKENIZER_DEC: "wavtokenizer-dec",
MODEL_ARCH.PLM: "plm",
MODEL_ARCH.BAILINGMOE: "bailingmoe",
MODEL_ARCH.DOTS1: "dots1",
MODEL_ARCH.ARCEE: "arcee",
}
VISION_PROJECTOR_TYPE_NAMES: dict[VISION_PROJECTOR_TYPE, str] = {
VISION_PROJECTOR_TYPE.MLP: "mlp",
VISION_PROJECTOR_TYPE.LDP: "ldp",
VISION_PROJECTOR_TYPE.LDPV2: "ldpv2",
VISION_PROJECTOR_TYPE.RESAMPLER: "resampler",
VISION_PROJECTOR_TYPE.GLM_EDGE: "adapter",
VISION_PROJECTOR_TYPE.MERGER: "qwen2vl_merger",
VISION_PROJECTOR_TYPE.GEMMA3: "gemma3",
}
TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
@@ -597,6 +734,8 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
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_B: "blk.{bid}.attn_kv_b",
MODEL_TENSOR.ATTN_K_B: "blk.{bid}.attn_k_b",
MODEL_TENSOR.ATTN_V_B: "blk.{bid}.attn_v_b",
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_SUB_NORM: "blk.{bid}.attn_sub_norm",
@@ -647,9 +786,126 @@ TENSOR_NAMES: dict[MODEL_TENSOR, str] = {
MODEL_TENSOR.POSNET_ATTN_K: "posnet.{bid}.attn_k",
MODEL_TENSOR.POSNET_ATTN_V: "posnet.{bid}.attn_v",
MODEL_TENSOR.POSNET_ATTN_OUT: "posnet.{bid}.attn_output",
# vision
MODEL_TENSOR.V_MMPROJ: "mm.{bid}",
MODEL_TENSOR.V_MMPROJ_FC: "mm.model.fc",
MODEL_TENSOR.V_MMPROJ_MLP: "mm.model.mlp.{bid}",
MODEL_TENSOR.V_MMPROJ_PEG: "mm.model.peg.{bid}",
MODEL_TENSOR.V_ENC_EMBD_CLS: "v.class_embd",
MODEL_TENSOR.V_ENC_EMBD_PATCH: "v.patch_embd",
MODEL_TENSOR.V_ENC_EMBD_POS: "v.position_embd",
MODEL_TENSOR.V_ENC_ATTN_Q: "v.blk.{bid}.attn_q",
MODEL_TENSOR.V_ENC_ATTN_Q_NORM: "v.blk.{bid}.attn_q_norm",
MODEL_TENSOR.V_ENC_ATTN_K: "v.blk.{bid}.attn_k",
MODEL_TENSOR.V_ENC_ATTN_K_NORM: "v.blk.{bid}.attn_k_norm",
MODEL_TENSOR.V_ENC_ATTN_V: "v.blk.{bid}.attn_v",
MODEL_TENSOR.V_ENC_INPUT_NORM: "v.blk.{bid}.ln1",
MODEL_TENSOR.V_ENC_ATTN_O: "v.blk.{bid}.attn_out",
MODEL_TENSOR.V_ENC_ATTN_O_NORM: "v.blk.{bid}.attn_out_norm",
MODEL_TENSOR.V_ENC_POST_ATTN_NORM: "v.blk.{bid}.ln2",
MODEL_TENSOR.V_ENC_FFN_UP: "v.blk.{bid}.ffn_up",
MODEL_TENSOR.V_ENC_FFN_GATE: "v.blk.{bid}.ffn_gate",
MODEL_TENSOR.V_ENC_FFN_DOWN: "v.blk.{bid}.ffn_down",
MODEL_TENSOR.V_LAYER_SCALE_1: "v.blk.{bid}.ls1",
MODEL_TENSOR.V_LAYER_SCALE_2: "v.blk.{bid}.ls2",
MODEL_TENSOR.V_PRE_NORM: "v.pre_ln",
MODEL_TENSOR.V_POST_NORM: "v.post_ln",
MODEL_TENSOR.V_MM_INP_PROJ: "mm.input_projection",
MODEL_TENSOR.V_MM_INP_NORM: "mm.input_norm",
MODEL_TENSOR.V_MM_SOFT_EMB_NORM: "mm.soft_emb_norm",
MODEL_TENSOR.V_RESMPL_POS_EMBD_K: "resampler.pos_embd_k",
MODEL_TENSOR.V_RESMPL_ATTN_Q: "resampler.attn.q",
MODEL_TENSOR.V_RESMPL_ATTN_K: "resampler.attn.k",
MODEL_TENSOR.V_RESMPL_ATTN_V: "resampler.attn.v",
MODEL_TENSOR.V_RESMPL_ATTN_OUT: "resampler.attn.out",
MODEL_TENSOR.V_RESMPL_KV: "resampler.kv",
MODEL_TENSOR.V_RESMPL_KV_NORM: "resampler.ln_kv",
MODEL_TENSOR.V_RESMPL_POST_NORM: "resampler.ln_post",
MODEL_TENSOR.V_RESMPL_Q_NORM: "resampler.ln_q",
MODEL_TENSOR.V_RESMPL_PROJ: "resampler.proj",
MODEL_TENSOR.V_RESMPL_QUERY: "resampler.query",
MODEL_TENSOR.V_TOK_EMBD_IMG_BREAK: "v.token_embd.img_break", # pixtral
MODEL_TENSOR.V_MM_PATCH_MERGER: "mm.patch_merger", # mistral small 3.1
# audio (mtmd)
MODEL_TENSOR.A_ENC_EMBD_POS: "a.position_embd",
MODEL_TENSOR.A_ENC_CONV1D: "a.conv1d.{bid}",
MODEL_TENSOR.A_PRE_NORM: "a.pre_ln",
MODEL_TENSOR.A_POST_NORM: "a.post_ln",
MODEL_TENSOR.A_ENC_ATTN_Q: "a.blk.{bid}.attn_q",
MODEL_TENSOR.A_ENC_ATTN_K: "a.blk.{bid}.attn_k",
MODEL_TENSOR.A_ENC_ATTN_V: "a.blk.{bid}.attn_v",
MODEL_TENSOR.A_ENC_INPUT_NORM: "a.blk.{bid}.ln1",
MODEL_TENSOR.A_ENC_OUTPUT: "a.blk.{bid}.attn_out",
MODEL_TENSOR.A_ENC_OUTPUT_NORM: "a.blk.{bid}.ln2",
MODEL_TENSOR.A_ENC_FFN_UP: "a.blk.{bid}.ffn_up",
MODEL_TENSOR.A_ENC_FFN_GATE: "a.blk.{bid}.ffn_gate",
MODEL_TENSOR.A_ENC_FFN_DOWN: "a.blk.{bid}.ffn_down",
MODEL_TENSOR.A_MMPROJ: "mm.a.mlp.{bid}",
MODEL_TENSOR.A_MMPROJ_FC: "mm.a.fc",
MODEL_TENSOR.A_MM_NORM_PRE: "mm.a.norm_pre",
MODEL_TENSOR.A_MM_NORM_MID: "mm.a.norm_mid",
}
MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_ARCH.MMPROJ: [
MODEL_TENSOR.V_MMPROJ,
MODEL_TENSOR.V_MMPROJ_FC,
MODEL_TENSOR.V_MMPROJ_MLP,
MODEL_TENSOR.V_MMPROJ_PEG,
MODEL_TENSOR.V_ENC_EMBD_CLS,
MODEL_TENSOR.V_ENC_EMBD_PATCH,
MODEL_TENSOR.V_ENC_EMBD_POS,
MODEL_TENSOR.V_ENC_INPUT_NORM,
MODEL_TENSOR.V_ENC_ATTN_Q,
MODEL_TENSOR.V_ENC_ATTN_Q_NORM,
MODEL_TENSOR.V_ENC_ATTN_K,
MODEL_TENSOR.V_ENC_ATTN_K_NORM,
MODEL_TENSOR.V_ENC_ATTN_V,
MODEL_TENSOR.V_ENC_ATTN_O,
MODEL_TENSOR.V_ENC_ATTN_O_NORM,
MODEL_TENSOR.V_ENC_POST_ATTN_NORM,
MODEL_TENSOR.V_ENC_FFN_UP,
MODEL_TENSOR.V_ENC_FFN_GATE,
MODEL_TENSOR.V_ENC_FFN_DOWN,
MODEL_TENSOR.V_LAYER_SCALE_1,
MODEL_TENSOR.V_LAYER_SCALE_2,
MODEL_TENSOR.V_PRE_NORM,
MODEL_TENSOR.V_POST_NORM,
MODEL_TENSOR.V_MM_INP_PROJ,
MODEL_TENSOR.V_MM_INP_NORM,
MODEL_TENSOR.V_MM_SOFT_EMB_NORM,
MODEL_TENSOR.V_RESMPL_POS_EMBD_K,
MODEL_TENSOR.V_RESMPL_ATTN_Q,
MODEL_TENSOR.V_RESMPL_ATTN_K,
MODEL_TENSOR.V_RESMPL_ATTN_V,
MODEL_TENSOR.V_RESMPL_ATTN_OUT,
MODEL_TENSOR.V_RESMPL_KV,
MODEL_TENSOR.V_RESMPL_KV_NORM,
MODEL_TENSOR.V_RESMPL_POST_NORM,
MODEL_TENSOR.V_RESMPL_Q_NORM,
MODEL_TENSOR.V_RESMPL_PROJ,
MODEL_TENSOR.V_RESMPL_QUERY,
MODEL_TENSOR.V_TOK_EMBD_IMG_BREAK,
MODEL_TENSOR.V_MM_PATCH_MERGER,
# audio
MODEL_TENSOR.A_ENC_EMBD_POS,
MODEL_TENSOR.A_ENC_CONV1D,
MODEL_TENSOR.A_PRE_NORM,
MODEL_TENSOR.A_POST_NORM,
MODEL_TENSOR.A_ENC_ATTN_Q,
MODEL_TENSOR.A_ENC_ATTN_K,
MODEL_TENSOR.A_ENC_ATTN_V,
MODEL_TENSOR.A_ENC_INPUT_NORM,
MODEL_TENSOR.A_ENC_OUTPUT,
MODEL_TENSOR.A_ENC_OUTPUT_NORM,
MODEL_TENSOR.A_ENC_FFN_UP,
MODEL_TENSOR.A_ENC_FFN_GATE,
MODEL_TENSOR.A_ENC_FFN_DOWN,
MODEL_TENSOR.A_MMPROJ,
MODEL_TENSOR.A_MMPROJ_FC,
MODEL_TENSOR.A_MM_NORM_PRE,
MODEL_TENSOR.A_MM_NORM_MID,
],
MODEL_ARCH.LLAMA: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
@@ -792,6 +1048,7 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.POS_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.ATTN_OUT_NORM,
MODEL_TENSOR.ATTN_QKV,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_V,
@@ -816,6 +1073,34 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.LAYER_OUT_NORM,
],
MODEL_ARCH.NOMIC_BERT_MOE: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.TOKEN_EMBD_NORM,
MODEL_TENSOR.TOKEN_TYPES,
MODEL_TENSOR.POS_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.ATTN_OUT_NORM,
MODEL_TENSOR.ATTN_QKV,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.FFN_GATE_INP,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
MODEL_TENSOR.LAYER_OUT_NORM,
],
MODEL_ARCH.NEO_BERT: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_QKV,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.ENC_OUTPUT_NORM,
MODEL_TENSOR.CLS,
MODEL_TENSOR.CLS_OUT,
],
MODEL_ARCH.JINA_BERT_V2: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.TOKEN_EMBD_NORM,
@@ -1524,6 +1809,8 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.ATTN_Q_B,
MODEL_TENSOR.ATTN_KV_A_MQA,
MODEL_TENSOR.ATTN_KV_B,
MODEL_TENSOR.ATTN_K_B,
MODEL_TENSOR.ATTN_V_B,
MODEL_TENSOR.ATTN_Q_A_NORM,
MODEL_TENSOR.ATTN_KV_A_NORM,
MODEL_TENSOR.ATTN_OUT,
@@ -1720,6 +2007,9 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_GATE_EXP,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_UP_EXP,
MODEL_TENSOR.FFN_GATE_SHEXP,
MODEL_TENSOR.FFN_UP_SHEXP,
MODEL_TENSOR.FFN_DOWN_SHEXP,
],
MODEL_ARCH.CHAMELEON: [
MODEL_TENSOR.TOKEN_EMBD,
@@ -1778,6 +2068,45 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = {
MODEL_TENSOR.FFN_DOWN_SHEXP,
MODEL_TENSOR.FFN_UP_SHEXP,
],
MODEL_ARCH.DOTS1: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_Q_NORM,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_K_NORM,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.FFN_EXP_PROBS_B,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_GATE,
MODEL_TENSOR.FFN_GATE_EXP,
MODEL_TENSOR.FFN_GATE_INP,
MODEL_TENSOR.FFN_GATE_SHEXP,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_DOWN_EXP,
MODEL_TENSOR.FFN_DOWN_SHEXP,
MODEL_TENSOR.FFN_UP,
MODEL_TENSOR.FFN_UP_EXP,
MODEL_TENSOR.FFN_UP_SHEXP,
],
MODEL_ARCH.ARCEE: [
MODEL_TENSOR.TOKEN_EMBD,
MODEL_TENSOR.OUTPUT_NORM,
MODEL_TENSOR.OUTPUT,
MODEL_TENSOR.ROPE_FREQS,
MODEL_TENSOR.ATTN_NORM,
MODEL_TENSOR.ATTN_Q,
MODEL_TENSOR.ATTN_K,
MODEL_TENSOR.ATTN_V,
MODEL_TENSOR.ATTN_OUT,
MODEL_TENSOR.ATTN_ROT_EMBD,
MODEL_TENSOR.FFN_NORM,
MODEL_TENSOR.FFN_DOWN,
MODEL_TENSOR.FFN_UP,
],
# TODO
}
@@ -1860,6 +2189,8 @@ class PoolingType(IntEnum):
NONE = 0
MEAN = 1
CLS = 2
LAST = 3
RANK = 4
class GGMLQuantizationType(IntEnum):
@@ -1986,6 +2317,19 @@ class GGUFValueType(IntEnum):
raise ValueError(f"Unknown type: {type(val)}")
class VisionProjectorType:
GEMMA3 = "gemma3"
IDEFICS3 = "idefics3"
PIXTRAL = "pixtral"
LLAMA4 = "llama4"
QWEN2VL = "qwen2vl_merger"
QWEN25VL = "qwen2.5vl_merger"
ULTRAVOX = "ultravox"
INTERNVL = "internvl"
QWEN2A = "qwen2a" # audio
QWEN25O = "qwen2.5o" # omni
# Items here are (block size, type size)
QK_K = 256
GGML_QUANT_SIZES: dict[GGMLQuantizationType, tuple[int, int]] = {