Gemma 2: ValueError: both hidden_act and hidden_activation are set

[agravier]

Describe the bug

I am not sure if that's a bug. Python3.10, M1.

from mistralrs import Runner, Which, ChatCompletionRequest, Architecture

runner = Runner(
    Which.Plain(
        model_id="google/gemma-2-9b-it",
        repeat_last_n=64,
        tokenizer_json=None,
        arch=Architecture.Gemma,
    )
)

⇒

2024-07-02T00:24:23.550226Z  INFO mistralrs_core::pipeline::normal: Loading `tokenizer.json` at `google/gemma-2-9b-it`
2024-07-02T00:24:23.551602Z  INFO mistralrs_core::pipeline::normal: Loading `config.json` at `google/gemma-2-9b-it`
2024-07-02T00:24:24.112827Z  INFO mistralrs_core::pipeline::normal: Loading `generation_config.json` at `google/gemma-2-9b-it`
2024-07-02T00:24:24.612251Z  INFO mistralrs_core::pipeline::normal: Loading `tokenizer_config.json` at `google/gemma-2-9b-it`
2024-07-02T00:24:24.618411Z  INFO mistralrs_core::utils::normal: DType selected is F16.
2024-07-02T00:24:24.618808Z  INFO mistralrs_core::pipeline::normal: Loading model `google/gemma-2-9b-it` on cpu.
2024-07-02T00:24:24.620317Z  INFO mistralrs_core::pipeline::normal: Model config: Config { attention_bias: false, head_dim: 256, hidden_act: Some(GeluPytorchTanh), hidden_activation: Some(GeluPytorchTanh), hidden_size: 3584, intermediate_size: 14336, num_attention_heads: 16, num_hidden_layers: 42, num_key_value_heads: 8, rms_norm_eps: 1e-6, rope_theta: 10000.0, vocab_size: 256000, max_position_embeddings: 8192, use_flash_attn: false }
2024-07-02T00:25:25.335118Z  INFO mistralrs_core::utils::normal: DType selected is F16.
Traceback (most recent call last):
  File "/Users/agravier/projects/perso/arc/archi/tests/mistral.py", line 21, in <module>
    runner = Runner(
ValueError: both hidden_act and hidden_activation are set

Latest commit

mistralrs-metal==0.1.24, which should be commit 3198004