Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[minor] typo & comments #441

Merged
merged 3 commits into from
Nov 18, 2024
Merged

[minor] typo & comments #441

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4925790
typo & comment
casinca Nov 17, 2024
7292733
comment
casinca Nov 17, 2024
5860057
update memory wording
rasbt Nov 18, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions ch05/07_gpt_to_llama/converting-gpt-to-llama2.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -381,7 +381,7 @@
"id": "qcD8LSHNhBRW"
},
"source": [
"- Note that we also added a `dtype=cfg[\"dtype\"]` setting above, which will allow us to load the model directly in lower precision formats later to save memory (versus instantiating it in the original 32-bit precision format and then converting it)\n",
"- Note that we also added a `dtype=cfg[\"dtype\"]` setting above, which will allow us to load the model directly in lower precision formats later to reduce memory usage (versus instantiating it in the original 32-bit precision format and then converting it)\n",
"- We also set `bias=False` since Llama doesn't use any bias units"
]
},
Expand Down Expand Up @@ -648,7 +648,7 @@
"\n",
"mha(example_batch)\n",
"\n",
"del mha # delete to safe memory"
"del mha # delete to free up memory"
]
},
{
Expand Down Expand Up @@ -781,7 +781,7 @@
" self.out_head = nn.Linear(cfg[\"emb_dim\"], cfg[\"vocab_size\"], bias=False, dtype=cfg[\"dtype\"])\n",
"\n",
" def forward(self, in_idx):\n",
" batch_size, seq_len = in_idx.shape\n",
" # batch_size, seq_len = in_idx.shape\n",
" tok_embeds = self.tok_emb(in_idx)\n",
" # pos_embeds = self.pos_emb(torch.arange(seq_len, device=in_idx.device))\n",
" x = tok_embeds # + pos_embeds # Shape [batch_size, num_tokens, emb_size]\n",
Expand Down Expand Up @@ -890,7 +890,7 @@
" \"n_heads\": 32, # Number of attention heads\n",
" \"n_layers\": 32, # Number of layers\n",
" \"hidden_dim\": 11008, # NEW: Size of the intermediate dimension in FeedForward\n",
" \"dtype\": torch.bfloat16 # NEW: Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16 # NEW: Lower-precision dtype to reduce memory usage\n",
"}"
]
},
Expand Down Expand Up @@ -1691,7 +1691,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.6"
"version": "3.11.4"
},
"widgets": {
"application/vnd.jupyter.widget-state+json": {
Expand Down
14 changes: 7 additions & 7 deletions ch05/07_gpt_to_llama/converting-llama2-to-llama3.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -481,7 +481,7 @@
" ):\n",
" super().__init__()\n",
" assert d_out % num_heads == 0, \"d_out must be divisible by num_heads\"\n",
" assert num_heads % num_kv_groups == 0, \"num_heads must be divisible by num_kv_groups\"\n",
" assert num_heads % num_kv_groups == 0, \"num_heads must be divisible by num_kv_groups\" # NEW\n",
"\n",
" self.d_out = d_out\n",
" self.num_heads = num_heads\n",
Expand Down Expand Up @@ -886,7 +886,7 @@
" \"n_heads\": 32, # Number of attention heads\n",
" \"n_layers\": 32, # Number of layers\n",
" \"hidden_dim\": 11_008, # Size of the intermediate dimension in FeedForward\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to reduce memory usage\n",
"}"
]
},
Expand All @@ -909,7 +909,7 @@
" \"n_kv_groups\": 8, # NEW: Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # NEW: The base in RoPE's \"theta\" was increased to 500_000\n",
" \"rope_freq\": None, # NEW: Additional configuration for adjusting the RoPE frequencies\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to reduce memory usage\n",
"}"
]
},
Expand Down Expand Up @@ -2062,7 +2062,7 @@
" \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
" \"rope_freq\": None, # Additional configuration for adjusting the RoPE frequencies\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16 # Lower-precision dtype to reduce memory usage\n",
"}\n",
"\n",
"LLAMA31_CONFIG_8B = {\n",
Expand All @@ -2074,7 +2074,7 @@
" \"hidden_dim\": 14_336, # Size of the intermediate dimension in FeedForward\n",
" \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to reduce memory usage\n",
" \"rope_freq\": { # NEW: RoPE frequency scaling\n",
" \"factor\": 8.0,\n",
" \"low_freq_factor\": 1.0,\n",
Expand Down Expand Up @@ -2448,7 +2448,7 @@
" \"hidden_dim\": 14_336, # Size of the intermediate dimension in FeedForward\n",
" \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to reduce memory usagey\n",
" \"rope_freq\": { # NEW: RoPE frequency scaling\n",
" \"factor\": 8.0,\n",
" \"low_freq_factor\": 1.0,\n",
Expand All @@ -2467,7 +2467,7 @@
" \"hidden_dim\": 8192, # NEW: Almost half the size of the intermediate dimension in FeedForward\n",
" \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to reduce memory usage\n",
" \"rope_freq\": { # RoPE frequency scaling\n",
" \"factor\": 32.0, # NEW: Adjustment of the rescaling factor\n",
" \"low_freq_factor\": 1.0,\n",
Expand Down
4 changes: 2 additions & 2 deletions ch05/07_gpt_to_llama/standalone-llama32.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -438,7 +438,7 @@
" \"hidden_dim\": 8192, # Size of the intermediate dimension in FeedForward\n",
" \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
" \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to save memory\n",
" \"dtype\": torch.bfloat16, # Lower-precision dtype to reduce memory usage\n",
" \"rope_freq\": { # RoPE frequency scaling\n",
" \"factor\": 32.0,\n",
" \"low_freq_factor\": 1.0,\n",
Expand All @@ -458,7 +458,7 @@
"# \"hidden_dim\": 8192, # Size of the intermediate dimension in FeedForward\n",
"# \"n_kv_groups\": 8, # Key-Value groups for grouped-query attention\n",
"# \"rope_base\": 500_000.0, # The base in RoPE's \"theta\"\n",
"# \"dtype\": torch.bfloat16, # Lower-precision dtype to save memory\n",
"# \"dtype\": torch.bfloat16, # Lower-precision dtype to reduce memory usage\n",
"# \"rope_freq\": { # RoPE frequency scaling\n",
"# \"factor\": 32.0,\n",
"# \"low_freq_factor\": 1.0,\n",
Expand Down