diff options
| author | liuwei-git <14815172+liuwei-git@users.noreply.github.com> | 2024-04-24 15:00:37 +0800 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-04-24 10:00:37 +0300 |
| commit | c8297c6af5693555652c40b95974b95d49d2674d (patch) | |
| tree | 403d5e80233b90b3a03a7283c297c679400ccf6a /gguf-py | |
| parent | 4e96a812b3ce7322a29a3008db2ed73d9087b176 (diff) | |
llama : add phi3 support (#6852)
* add explicit phi3 support
* add explicit phi3 support
* remove unused code
* convert : add BOS token
* llama : match EOT token <|end|>
* llama : minor / style
* llama : tabs -> spaces
* convert : fix lint checks
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Diffstat (limited to 'gguf-py')
| -rw-r--r-- | gguf-py/gguf/constants.py | 16 | ||||
| -rw-r--r-- | gguf-py/gguf/tensor_mapping.py | 2 |
2 files changed, 18 insertions, 0 deletions
diff --git a/gguf-py/gguf/constants.py b/gguf-py/gguf/constants.py index 06cb26a7..d2f1de19 100644 --- a/gguf-py/gguf/constants.py +++ b/gguf-py/gguf/constants.py @@ -124,6 +124,7 @@ class MODEL_ARCH(IntEnum): QWEN2 = auto() QWEN2MOE = auto() PHI2 = auto() + PHI3 = auto() PLAMO = auto() CODESHELL = auto() ORION = auto() @@ -200,6 +201,7 @@ MODEL_ARCH_NAMES: dict[MODEL_ARCH, str] = { MODEL_ARCH.QWEN2: "qwen2", MODEL_ARCH.QWEN2MOE: "qwen2moe", MODEL_ARCH.PHI2: "phi2", + MODEL_ARCH.PHI3: "phi3", MODEL_ARCH.PLAMO: "plamo", MODEL_ARCH.CODESHELL: "codeshell", MODEL_ARCH.ORION: "orion", @@ -550,6 +552,20 @@ MODEL_TENSORS: dict[MODEL_ARCH, list[MODEL_TENSOR]] = { MODEL_TENSOR.FFN_DOWN, MODEL_TENSOR.FFN_UP, ], + MODEL_ARCH.PHI3: [ + MODEL_TENSOR.TOKEN_EMBD, + MODEL_TENSOR.OUTPUT_NORM, + MODEL_TENSOR.OUTPUT, + MODEL_TENSOR.ATTN_NORM, + MODEL_TENSOR.ATTN_QKV, + MODEL_TENSOR.ATTN_Q, + MODEL_TENSOR.ATTN_K, + MODEL_TENSOR.ATTN_V, + MODEL_TENSOR.ATTN_OUT, + MODEL_TENSOR.FFN_NORM, + MODEL_TENSOR.FFN_DOWN, + MODEL_TENSOR.FFN_UP, + ], MODEL_ARCH.CODESHELL: [ MODEL_TENSOR.TOKEN_EMBD, MODEL_TENSOR.POS_EMBD, diff --git a/gguf-py/gguf/tensor_mapping.py b/gguf-py/gguf/tensor_mapping.py index 10de36fa..e5750d41 100644 --- a/gguf-py/gguf/tensor_mapping.py +++ b/gguf-py/gguf/tensor_mapping.py @@ -117,6 +117,7 @@ class TensorNameMap: "h.{bid}.attn.c_attn", # gpt2 "transformer.h.{bid}.mixer.Wqkv", # phi2 "encoder.layers.{bid}.attn.Wqkv", # nomic-bert + "model.layers.{bid}.self_attn.qkv_proj" # phi3 ), # Attention query @@ -234,6 +235,7 @@ class TensorNameMap: "h.{bid}.mlp.c_fc", # gpt2 "transformer.h.{bid}.mlp.fc1", # phi2 "model.layers.{bid}.mlp.fc1", # phi2 + "model.layers.{bid}.mlp.gate_up_proj", # phi3 "model.layers.layers.{bid}.mlp.up_proj", # plamo "model.layers.{bid}.feed_forward.w3", # internlm2 "encoder.layers.{bid}.mlp.fc11", # nomic-bert |