mirror of
https://github.com/rasbt/LLMs-from-scratch.git
synced 2026-04-11 14:21:41 +08:00
update
This commit is contained in:
parent
341ff65beb
commit
ea8cf1fbc3
1
.github/workflows/basic-tests-linux-uv.yml
vendored
1
.github/workflows/basic-tests-linux-uv.yml
vendored
@ -79,5 +79,4 @@ jobs:
|
||||
shell: bash
|
||||
run: |
|
||||
source .venv/bin/activate
|
||||
uv pip install transformers -U
|
||||
pytest pkg/llms_from_scratch/tests/
|
||||
|
||||
1
.github/workflows/basic-tests-macos-uv.yml
vendored
1
.github/workflows/basic-tests-macos-uv.yml
vendored
@ -47,7 +47,6 @@ jobs:
|
||||
shell: bash
|
||||
run: |
|
||||
source .venv/bin/activate
|
||||
uv pip install transformers -U
|
||||
pytest setup/02_installing-python-libraries/tests.py
|
||||
pytest ch04/01_main-chapter-code/tests.py
|
||||
pytest ch05/01_main-chapter-code/tests.py
|
||||
|
||||
@ -49,7 +49,6 @@ jobs:
|
||||
shell: bash
|
||||
run: |
|
||||
source .venv/Scripts/activate
|
||||
uv pip install transformers -U
|
||||
pytest setup/02_installing-python-libraries/tests.py
|
||||
pytest ch04/01_main-chapter-code/tests.py
|
||||
pytest ch05/01_main-chapter-code/tests.py
|
||||
|
||||
@ -177,6 +177,10 @@ def test_rope_llama2(notebook):
|
||||
max_position_embeddings: int = 8192
|
||||
hidden_size = head_dim * num_heads
|
||||
num_attention_heads = num_heads
|
||||
rope_parameters = {"rope_type": "default", "rope_theta": theta_base}
|
||||
|
||||
def standardize_rope_params(self):
|
||||
return
|
||||
|
||||
config = RoPEConfig()
|
||||
rot_emb = LlamaRotaryEmbedding(config=config)
|
||||
@ -242,6 +246,10 @@ def test_rope_llama3(notebook):
|
||||
max_position_embeddings: int = 8192
|
||||
hidden_size = head_dim * num_heads
|
||||
num_attention_heads = num_heads
|
||||
rope_parameters = {"rope_type": "default", "rope_theta": theta_base}
|
||||
|
||||
def standardize_rope_params(self):
|
||||
return
|
||||
|
||||
config = RoPEConfig()
|
||||
rot_emb = LlamaRotaryEmbedding(config=config)
|
||||
@ -320,6 +328,10 @@ def test_rope_llama3_12(notebook):
|
||||
max_position_embeddings: int = 8192
|
||||
hidden_size = head_dim * num_heads
|
||||
num_attention_heads = num_heads
|
||||
rope_parameters = {**hf_rope_params, "rope_theta": rope_theta}
|
||||
|
||||
def standardize_rope_params(self):
|
||||
return
|
||||
|
||||
config = RoPEConfig()
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user