code stringlengths 3 6.57k |
|---|
torch.tensor(0.) |
len(grads) |
torch.norm(grads[0], p=2, dtype=torch.float32) |
multi_tensor_total_norm(grads) |
torch.stack([torch.norm(g, p=2, dtype=torch.float32) |
aggregate_norm_fn(total_norm) |
float(max_norm) |
clamp_(max=1) |
g.mul_(clip_coef) |
fill_with_neg_inf(t) |
t.float() |
fill_(float("-inf") |
type_as(t) |
_match_types(arg1, arg2) |
upgrade(arg_number, arg_structure) |
isinstance(arg_structure, tuple) |
tuple([arg_number] * len(arg_structure) |
isinstance(arg_structure, dict) |
copy.deepcopy(arg_structure) |
upgrade(arg_number, arg_structure[k]) |
isinstance(arg1, float) |
isinstance(arg1, int) |
upgrade(arg1, arg2) |
isinstance(arg2, float) |
isinstance(arg2, int) |
upgrade(arg2, arg1) |
resolve_max_positions(*args) |
map_value_update(d1, d2) |
copy.deepcopy(d1) |
min(d1[key], d2[key]) |
nullsafe_min(l) |
_match_types(max_positions, arg) |
isinstance(arg, float) |
isinstance(arg, int) |
min(max_positions, arg) |
isinstance(arg, dict) |
map_value_update(max_positions, arg) |
tuple(map(nullsafe_min, zip(max_positions, arg) |
import_user_module(args) |
getattr(args, "user_dir", None) |
os.path.abspath(args.user_dir) |
os.path.exists(module_path) |
os.path.dirname(__file__) |
os.path.exists(fairseq_rel_path) |
os.path.split(module_path) |
sys.path.insert(0, module_parent) |
importlib.import_module(module_name) |
softmax(x, dim: int, onnx_trace: bool = False) |
F.softmax(x.float() |
F.softmax(x, dim=dim, dtype=torch.float32) |
log_softmax(x, dim: int, onnx_trace: bool = False) |
F.log_softmax(x.float() |
F.log_softmax(x, dim=dim, dtype=torch.float32) |
get_perplexity(loss, round=2, base=2) |
safe_round(base ** loss, round) |
float('inf') |
deprecation_warning(message, stacklevel=3) |
warnings.warn(message, stacklevel=stacklevel) |
get_activation_fn(activation: str) |
RuntimeError("--activation-fn {} not supported".format(activation) |
get_available_activation_fns() |
eval(model) |
model.eval() |
model.train(is_training) |
has_parameters(module) |
next(module.parameters() |
set_torch_seed(seed) |
isinstance(seed, int) |
torch.manual_seed(seed) |
torch.cuda.manual_seed(seed) |
with_torch_seed(seed) |
isinstance(seed, int) |
torch.get_rng_state() |
torch.cuda.get_rng_state() |
set_torch_seed(seed) |
torch.set_rng_state(rng_state) |
torch.cuda.set_rng_state(cuda_rng_state) |
parse_alignment(line) |
line (str) |
shape (2 * m) |
line.strip() |
split() |
torch.IntTensor(2 * len(alignments) |
enumerate(alignments) |
alignment.split("-") |
int(src_idx) |
int(tgt_idx) |
get_token_to_word_mapping(tokens, exclude_list) |
len(tokens) |
int(token not in exclude_list) |
list(accumulate(word_start) |
range(n) |
extract_hard_alignment(attn, src_sent, tgt_sent, pad, eos) |
nonzero() |
squeeze(dim=-1) |
nonzero() |
squeeze(dim=-1) |
get_token_to_word_mapping(src_sent, [eos, pad]) |
get_token_to_word_mapping(tgt_sent, [eos, pad]) |
len(tgt_valid) |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.