juanwisz/modernbert-python-code-retrieval
0105
1---2tags:3- sentence-transformers4- sentence-similarity5- feature-extraction6- generated_from_trainer7- dataset_size:4121788- loss:MultipleNegativesRankingLoss9base_model: answerdotai/ModernBERT-base10widget:11- source_sentence: "Clip off all parts from all bounding boxes that are outside of\12 \ the image.\n\n Returns\n -------\n imgaug.BoundingBoxesOnImage\n\13 \ Bounding boxes, clipped to fall within the image dimensions."14 sentences:15 - "def model_best(y1, y2, samples=1000, progressbar=True):\n \"\"\"\n Bayesian\16 \ Estimation Supersedes the T-Test\n\n This model runs a Bayesian hypothesis\17 \ comparing if y1 and y2 come\n from the same distribution. Returns are assumed\18 \ to be T-distributed.\n\n In addition, computes annual volatility and Sharpe\19 \ of in and\n out-of-sample periods.\n\n This model replicates the example\20 \ used in:\n Kruschke, John. (2012) Bayesian estimation supersedes the t\n\21 \ test. Journal of Experimental Psychology: General.\n\n Parameters\n \22 \ ----------\n y1 : array-like\n Array of returns (e.g. in-sample)\n\23 \ y2 : array-like\n Array of returns (e.g. out-of-sample)\n samples\24 \ : int, optional\n Number of posterior samples to draw.\n\n Returns\n\25 \ -------\n model : pymc.Model object\n PyMC3 model containing all\26 \ random variables.\n trace : pymc3.sampling.BaseTrace object\n A PyMC3\27 \ trace object that contains samples for each parameter\n of the posterior.\n\28 \n See Also\n --------\n plot_stoch_vol : plotting of tochastic volatility\29 \ model\n \"\"\"\n\n y = np.concatenate((y1, y2))\n\n mu_m = np.mean(y)\n\30 \ mu_p = 0.000001 * 1 / np.std(y)**2\n\n sigma_low = np.std(y) / 1000\n\31 \ sigma_high = np.std(y) * 1000\n with pm.Model() as model:\n group1_mean\32 \ = pm.Normal('group1_mean', mu=mu_m, tau=mu_p,\n \33 \ testval=y1.mean())\n group2_mean = pm.Normal('group2_mean', mu=mu_m,\34 \ tau=mu_p,\n testval=y2.mean())\n group1_std\35 \ = pm.Uniform('group1_std', lower=sigma_low,\n \36 \ upper=sigma_high, testval=y1.std())\n group2_std = pm.Uniform('group2_std',\37 \ lower=sigma_low,\n upper=sigma_high, testval=y2.std())\n\38 \ nu = pm.Exponential('nu_minus_two', 1 / 29., testval=4.) + 2.\n\n \39 \ returns_group1 = pm.StudentT('group1', nu=nu, mu=group1_mean,\n \40 \ lam=group1_std**-2, observed=y1)\n returns_group2\41 \ = pm.StudentT('group2', nu=nu, mu=group2_mean,\n \42 \ lam=group2_std**-2, observed=y2)\n\n diff_of_means = pm.Deterministic('difference\43 \ of means',\n group2_mean - group1_mean)\n\44 \ pm.Deterministic('difference of stds',\n group2_std\45 \ - group1_std)\n pm.Deterministic('effect size', diff_of_means /\n \46 \ pm.math.sqrt((group1_std**2 +\n \47 \ group2_std**2) / 2))\n\n pm.Deterministic('group1_annual_volatility',\n\48 \ returns_group1.distribution.variance**.5 *\n \49 \ np.sqrt(252))\n pm.Deterministic('group2_annual_volatility',\n\50 \ returns_group2.distribution.variance**.5 *\n \51 \ np.sqrt(252))\n\n pm.Deterministic('group1_sharpe',\52 \ returns_group1.distribution.mean /\n returns_group1.distribution.variance**.5\53 \ *\n np.sqrt(252))\n pm.Deterministic('group2_sharpe',\54 \ returns_group2.distribution.mean /\n returns_group2.distribution.variance**.5\55 \ *\n np.sqrt(252))\n\n trace = pm.sample(samples,\56 \ progressbar=progressbar)\n return model, trace"57 - "def clip_out_of_image(self):\n \"\"\"\n Clip off all parts from\58 \ all bounding boxes that are outside of the image.\n\n Returns\n \59 \ -------\n imgaug.BoundingBoxesOnImage\n Bounding boxes,\60 \ clipped to fall within the image dimensions.\n\n \"\"\"\n bbs_cut\61 \ = [bb.clip_out_of_image(self.shape)\n for bb in self.bounding_boxes\62 \ if bb.is_partly_within_image(self.shape)]\n return BoundingBoxesOnImage(bbs_cut,\63 \ shape=self.shape)"64 - "def _initPermanence(self, potential, connectedPct):\n \"\"\"\n Initializes\65 \ the permanences of a column. The method\n returns a 1-D array the size of\66 \ the input, where each entry in the\n array represents the initial permanence\67 \ value between the input bit\n at the particular index in the array, and the\68 \ column represented by\n the 'index' parameter.\n\n Parameters:\n ----------------------------\n\69 \ :param potential: A numpy array specifying the potential pool of the column.\n\70 \ Permanence values will only be generated for input bits\n\71 \ corresponding to indices for which the mask value is 1.\n\72 \ :param connectedPct: A value between 0 or 1 governing the chance, for each\n\73 \ permanence, that the initial permanence value will\n\74 \ be a value that is considered connected.\n \"\"\"\75 \n # Determine which inputs bits will start out as connected\n # to the\76 \ inputs. Initially a subset of the input bits in a\n # column's potential\77 \ pool will be connected. This number is\n # given by the parameter \"connectedPct\"\78 \n perm = numpy.zeros(self._numInputs, dtype=realDType)\n for i in xrange(self._numInputs):\n\79 \ if (potential[i] < 1):\n continue\n\n if (self._random.getReal64()\80 \ <= connectedPct):\n perm[i] = self._initPermConnected()\n else:\n\81 \ perm[i] = self._initPermNonConnected()\n\n # Clip off low values.\82 \ Since we use a sparse representation\n # to store the permanence values this\83 \ helps reduce memory\n # requirements.\n perm[perm < self._synPermTrimThreshold]\84 \ = 0\n\n return perm"85- source_sentence: "Perform a weighted average over dicts that are each on a different\86 \ node\n Input: local_name2valcount: dict mapping key -> (value, count)\n \87 \ Returns: key -> mean"88 sentences:89 - "def MotionBlur(k=5, angle=(0, 360), direction=(-1.0, 1.0), order=1, name=None,\90 \ deterministic=False, random_state=None):\n \"\"\"\n Augmenter that sharpens\91 \ images and overlays the result with the original image.\n\n dtype support::\n\92 \n See ``imgaug.augmenters.convolutional.Convolve``.\n\n Parameters\n\93 \ ----------\n k : int or tuple of int or list of int or imgaug.parameters.StochasticParameter,\94 \ optional\n Kernel size to use.\n\n * If a single int, then\95 \ that value will be used for the height\n and width of the kernel.\n\96 \ * If a tuple of two ints ``(a, b)``, then the kernel size will be\n\97 \ sampled from the interval ``[a..b]``.\n * If a list,\98 \ then a random value will be sampled from that list per image.\n *\99 \ If a StochasticParameter, then ``N`` samples will be drawn from\n \100 \ that parameter per ``N`` input images, each representing the kernel\n \101 \ size for the nth image.\n\n angle : number or tuple of number or\102 \ list of number or imgaug.parameters.StochasticParameter, optional\n Angle\103 \ of the motion blur in degrees (clockwise, relative to top center direction).\n\104 \n * If a number, exactly that value will be used.\n * If\105 \ a tuple ``(a, b)``, a random value from the range ``a <= x <= b`` will\n \106 \ be sampled per image.\n * If a list, then a random value\107 \ will be sampled from that list per image.\n * If a StochasticParameter,\108 \ a value will be sampled from the\n parameter per image.\n\n \109 \ direction : number or tuple of number or list of number or imgaug.parameters.StochasticParameter,\110 \ optional\n Forward/backward direction of the motion blur. Lower values\111 \ towards -1.0 will point the motion blur towards\n the back (with angle\112 \ provided via `angle`). Higher values towards 1.0 will point the motion blur\113 \ forward.\n A value of 0.0 leads to a uniformly (but still angled) motion\114 \ blur.\n\n * If a number, exactly that value will be used.\n \115 \ * If a tuple ``(a, b)``, a random value from the range ``a <= x <= b``\116 \ will\n be sampled per image.\n * If a list, then a random\117 \ value will be sampled from that list per image.\n * If a StochasticParameter,\118 \ a value will be sampled from the\n parameter per image.\n\n \119 \ order : int or iterable of int or imgaug.ALL or imgaug.parameters.StochasticParameter,\120 \ optional\n Interpolation order to use when rotating the kernel according\121 \ to `angle`.\n See :func:`imgaug.augmenters.geometric.Affine.__init__`.\n\122 \ Recommended to be ``0`` or ``1``, with ``0`` being faster, but less continuous/smooth\123 \ as `angle` is changed,\n particularly around multiple of 45 degrees.\n\124 \n name : None or str, optional\n See :func:`imgaug.augmenters.meta.Augmenter.__init__`.\n\125 \n deterministic : bool, optional\n See :func:`imgaug.augmenters.meta.Augmenter.__init__`.\n\126 \n random_state : None or int or numpy.random.RandomState, optional\n \127 \ See :func:`imgaug.augmenters.meta.Augmenter.__init__`.\n\n Examples\n \128 \ --------\n >>> aug = iaa.MotionBlur(k=15)\n\n Create a motion blur augmenter\129 \ with kernel size of 15x15.\n\n >>> aug = iaa.MotionBlur(k=15, angle=[-45,\130 \ 45])\n\n Create a motion blur augmenter with kernel size of 15x15 and a blur\131 \ angle of either -45 or 45 degrees (randomly\n picked per image).\n\n \"\132 \"\"\n # TODO allow (1, None) and set to identity matrix if k == 1\n k_param\133 \ = iap.handle_discrete_param(k, \"k\", value_range=(3, None), tuple_to_uniform=True,\134 \ list_to_choice=True,\n allow_floats=False)\n\135 \ angle_param = iap.handle_continuous_param(angle, \"angle\", value_range=None,\136 \ tuple_to_uniform=True,\n list_to_choice=True)\n\137 \ direction_param = iap.handle_continuous_param(direction, \"direction\", value_range=(-1.0-1e-6,\138 \ 1.0+1e-6),\n tuple_to_uniform=True,\139 \ list_to_choice=True)\n\n def create_matrices(image, nb_channels, random_state_func):\n\140 \ # avoid cyclic import between blur and geometric\n from . import\141 \ geometric as iaa_geometric\n\n # force discrete for k_sample via int()\142 \ in case of stochastic parameter\n k_sample = int(k_param.draw_sample(random_state=random_state_func))\n\143 \ angle_sample = angle_param.draw_sample(random_state=random_state_func)\n\144 \ direction_sample = direction_param.draw_sample(random_state=random_state_func)\n\145 \n k_sample = k_sample if k_sample % 2 != 0 else k_sample + 1\n \146 \ direction_sample = np.clip(direction_sample, -1.0, 1.0)\n direction_sample\147 \ = (direction_sample + 1.0) / 2.0\n\n matrix = np.zeros((k_sample, k_sample),\148 \ dtype=np.float32)\n matrix[:, k_sample//2] = np.linspace(float(direction_sample),\149 \ 1.0 - float(direction_sample), num=k_sample)\n rot = iaa_geometric.Affine(rotate=angle_sample,\150 \ order=order)\n matrix = (rot.augment_image((matrix * 255).astype(np.uint8))\151 \ / 255.0).astype(np.float32)\n\n return [matrix/np.sum(matrix)] * nb_channels\n\152 \n if name is None:\n name = \"Unnamed%s\" % (ia.caller_name(),)\n\n\153 \ return iaa_convolutional.Convolve(create_matrices, name=name, deterministic=deterministic,\n\154 \ random_state=random_state)"155 - "def rolling_sharpe(returns, rolling_sharpe_window):\n \"\"\"\n Determines\156 \ the rolling Sharpe ratio of a strategy.\n\n Parameters\n ----------\n\157 \ returns : pd.Series\n Daily returns of the strategy, noncumulative.\n\158 \ - See full explanation in tears.create_full_tear_sheet.\n rolling_sharpe_window\159 \ : int\n Length of rolling window, in days, over which to compute.\n\n\160 \ Returns\n -------\n pd.Series\n Rolling Sharpe ratio.\n\n \161 \ Note\n -----\n See https://en.wikipedia.org/wiki/Sharpe_ratio for more\162 \ details.\n \"\"\"\n\n return returns.rolling(rolling_sharpe_window).mean()\163 \ \\\n / returns.rolling(rolling_sharpe_window).std() \\\n * np.sqrt(APPROX_BDAYS_PER_YEAR)"164 - "def mpi_weighted_mean(comm, local_name2valcount):\n \"\"\"\n Perform a\165 \ weighted average over dicts that are each on a different node\n Input: local_name2valcount:\166 \ dict mapping key -> (value, count)\n Returns: key -> mean\n \"\"\"\n \167 \ all_name2valcount = comm.gather(local_name2valcount)\n if comm.rank ==\168 \ 0:\n name2sum = defaultdict(float)\n name2count = defaultdict(float)\n\169 \ for n2vc in all_name2valcount:\n for (name, (val, count))\170 \ in n2vc.items():\n try:\n val = float(val)\n\171 \ except ValueError:\n if comm.rank == 0:\n\172 \ warnings.warn('WARNING: tried to compute mean on non-float\173 \ {}={}'.format(name, val))\n else:\n name2sum[name]\174 \ += val * count\n name2count[name] += count\n return\175 \ {name : name2sum[name] / name2count[name] for name in name2sum}\n else:\n\176 \ return {}"177- source_sentence: "Generate and return the following encoder related substitution\178 \ variables:\n\n encoderSpecsStr:\n For the base description file, this string\179 \ defines the default\n encoding dicts for each encoder. For example:\n \180 \ '__gym_encoder' : { 'fieldname': 'gym',\n 'n': 13,\n \181 \ 'name': 'gym',\n 'type': 'SDRCategoryEncoder',\n 'w': 7},\n\182 \ '__address_encoder' : { 'fieldname': 'address',\n 'n': 13,\n\183 \ 'name': 'address',\n 'type': 'SDRCategoryEncoder',\n \184 \ 'w': 7}\n\n encoderSchemaStr:\n For the base description file, this\185 \ is a list containing a\n DeferredDictLookup entry for each encoder. For example:\n\186 \ [DeferredDictLookup('__gym_encoder'),\n DeferredDictLookup('__address_encoder'),\n\187 \ DeferredDictLookup('__timestamp_timeOfDay_encoder'),\n DeferredDictLookup('__timestamp_dayOfWeek_encoder'),\n\188 \ DeferredDictLookup('__consumption_encoder')],\n\n permEncoderChoicesStr:\n\189 \ For the permutations file, this defines the possible\n encoder dicts for\190 \ each encoder. For example:\n '__timestamp_dayOfWeek_encoder': [\n \191 \ None,\n {'fieldname':'timestamp',\n \192 \ 'name': 'timestamp_timeOfDay',\n 'type':'DateEncoder'\n\193 \ 'dayOfWeek': (7,1)\n },\n \194 \ {'fieldname':'timestamp',\n 'name': 'timestamp_timeOfDay',\n\195 \ 'type':'DateEncoder'\n 'dayOfWeek':\196 \ (7,3)\n },\n ],\n\n '__field_consumption_encoder':\197 \ [\n None,\n {'fieldname':'consumption',\n\198 \ 'name': 'consumption',\n 'type':'AdaptiveScalarEncoder',\n\199 \ 'n': 13,\n 'w': 7,\n \200 \ }\n ]\n\n\n\n Parameters:\n --------------------------------------------------\n\201 \ includedFields: item from the 'includedFields' section of the\n \202 \ description JSON object. This is a list of dicts, each\n \203 \ dict defining the field name, type, and optional min\n \204 \ and max values.\n\n retval: (encoderSpecsStr, encoderSchemaStr permEncoderChoicesStr)"205 sentences:206 - "def _generateEncoderStringsV1(includedFields):\n \"\"\" Generate and return\207 \ the following encoder related substitution variables:\n\n encoderSpecsStr:\n\208 \ For the base description file, this string defines the default\n encoding\209 \ dicts for each encoder. For example:\n '__gym_encoder' : { 'fieldname':\210 \ 'gym',\n 'n': 13,\n 'name': 'gym',\n 'type': 'SDRCategoryEncoder',\n\211 \ 'w': 7},\n '__address_encoder' : { 'fieldname': 'address',\n\212 \ 'n': 13,\n 'name': 'address',\n 'type': 'SDRCategoryEncoder',\n\213 \ 'w': 7}\n\n encoderSchemaStr:\n For the base description file,\214 \ this is a list containing a\n DeferredDictLookup entry for each encoder.\215 \ For example:\n [DeferredDictLookup('__gym_encoder'),\n DeferredDictLookup('__address_encoder'),\n\216 \ DeferredDictLookup('__timestamp_timeOfDay_encoder'),\n DeferredDictLookup('__timestamp_dayOfWeek_encoder'),\n\217 \ DeferredDictLookup('__consumption_encoder')],\n\n permEncoderChoicesStr:\n\218 \ For the permutations file, this defines the possible\n encoder dicts for\219 \ each encoder. For example:\n '__timestamp_dayOfWeek_encoder': [\n \220 \ None,\n {'fieldname':'timestamp',\n \221 \ 'name': 'timestamp_timeOfDay',\n 'type':'DateEncoder'\n\222 \ 'dayOfWeek': (7,1)\n },\n \223 \ {'fieldname':'timestamp',\n 'name': 'timestamp_timeOfDay',\n\224 \ 'type':'DateEncoder'\n 'dayOfWeek':\225 \ (7,3)\n },\n ],\n\n '__field_consumption_encoder':\226 \ [\n None,\n {'fieldname':'consumption',\n\227 \ 'name': 'consumption',\n 'type':'AdaptiveScalarEncoder',\n\228 \ 'n': 13,\n 'w': 7,\n \229 \ }\n ]\n\n\n\n Parameters:\n --------------------------------------------------\n\230 \ includedFields: item from the 'includedFields' section of the\n \231 \ description JSON object. This is a list of dicts, each\n \232 \ dict defining the field name, type, and optional min\n \233 \ and max values.\n\n retval: (encoderSpecsStr, encoderSchemaStr permEncoderChoicesStr)\n\234 \n\n \"\"\"\n\n # ------------------------------------------------------------------------\n\235 \ # First accumulate the possible choices for each encoder\n encoderChoicesList\236 \ = []\n for fieldInfo in includedFields:\n\n fieldName = fieldInfo['fieldName']\n\237 \n # Get the list of encoder choices for this field\n (choicesList, aggFunction)\238 \ = _generateEncoderChoicesV1(fieldInfo)\n encoderChoicesList.extend(choicesList)\n\239 \n\n # ------------------------------------------------------------------------\n\240 \ # Generate the string containing the encoder specs and encoder schema. See\n\241 \ # the function comments for an example of the encoderSpecsStr and\n # encoderSchemaStr\n\242 \ #\n encoderSpecsList = []\n for encoderChoices in encoderChoicesList:\n \243 \ # Use the last choice as the default in the base file because the 1st is\n\244 \ # often None\n encoder = encoderChoices[-1]\n\n # Check for bad characters\n\245 \ for c in _ILLEGAL_FIELDNAME_CHARACTERS:\n if encoder['name'].find(c)\246 \ >= 0:\n raise _ExpGeneratorException(\"Illegal character in field: %r\247 \ (%r)\" % (\n c, encoder['name']))\n\n encoderSpecsList.append(\"\248 %s: \\n%s%s\" % (\n _quoteAndEscape(encoder['name']),\n 2*_ONE_INDENT,\n\249 \ pprint.pformat(encoder, indent=2*_INDENT_STEP)))\n\n encoderSpecsStr\250 \ = ',\\n '.join(encoderSpecsList)\n\n\n # ------------------------------------------------------------------------\n\251 \ # Generate the string containing the permutation encoder choices. See the\n\252 \ # function comments above for an example of the permEncoderChoicesStr\n\n\253 \ permEncoderChoicesList = []\n for encoderChoices in encoderChoicesList:\n\254 \ permEncoderChoicesList.append(\"%s: %s,\" % (\n _quoteAndEscape(encoderChoices[-1]['name']),\n\255 \ pprint.pformat(encoderChoices, indent=2*_INDENT_STEP)))\n permEncoderChoicesStr\256 \ = '\\n'.join(permEncoderChoicesList)\n permEncoderChoicesStr = _indentLines(permEncoderChoicesStr,\257 \ 1,\n indentFirstLine=False)\n\n # Return\258 \ results\n return (encoderSpecsStr, permEncoderChoicesStr)"259 - "def shift(self, top=None, right=None, bottom=None, left=None):\n \"\"\"\260 \n Shift/move the line strings from one or more image sides.\n\n \261 \ Parameters\n ----------\n top : None or int, optional\n \262 \ Amount of pixels by which to shift all bounding boxes from the\n \263 \ top.\n\n right : None or int, optional\n Amount of pixels\264 \ by which to shift all bounding boxes from the\n right.\n\n \265 \ bottom : None or int, optional\n Amount of pixels by which to shift\266 \ all bounding boxes from the\n bottom.\n\n left : None or int,\267 \ optional\n Amount of pixels by which to shift all bounding boxes\268 \ from the\n left.\n\n Returns\n -------\n imgaug.augmentables.lines.LineStringsOnImage\n\269 \ Shifted line strings.\n\n \"\"\"\n lss_new = [ls.shift(top=top,\270 \ right=right, bottom=bottom, left=left)\n for ls in self.line_strings]\n\271 \ return LineStringsOnImage(lss_new, shape=self.shape)"272 - "def cross_entropy_reward_loss(logits, actions, rewards, name=None):\n \"\"\273 \"Calculate the loss for Policy Gradient Network.\n\n Parameters\n ----------\n\274 \ logits : tensor\n The network outputs without softmax. This function\275 \ implements softmax inside.\n actions : tensor or placeholder\n The\276 \ agent actions.\n rewards : tensor or placeholder\n The rewards.\n\n\277 \ Returns\n --------\n Tensor\n The TensorFlow loss function.\n\278 \n Examples\n ----------\n >>> states_batch_pl = tf.placeholder(tf.float32,\279 \ shape=[None, D])\n >>> network = InputLayer(states_batch_pl, name='input')\n\280 \ >>> network = DenseLayer(network, n_units=H, act=tf.nn.relu, name='relu1')\n\281 \ >>> network = DenseLayer(network, n_units=3, name='out')\n >>> probs =\282 \ network.outputs\n >>> sampling_prob = tf.nn.softmax(probs)\n >>> actions_batch_pl\283 \ = tf.placeholder(tf.int32, shape=[None])\n >>> discount_rewards_batch_pl\284 \ = tf.placeholder(tf.float32, shape=[None])\n >>> loss = tl.rein.cross_entropy_reward_loss(probs,\285 \ actions_batch_pl, discount_rewards_batch_pl)\n >>> train_op = tf.train.RMSPropOptimizer(learning_rate,\286 \ decay_rate).minimize(loss)\n\n \"\"\"\n cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(labels=actions,\287 \ logits=logits, name=name)\n\n return tf.reduce_sum(tf.multiply(cross_entropy,\288 \ rewards))"289- source_sentence: "Translate an index into coordinates, using the given coordinate\290 \ system.\n\n Similar to ``numpy.unravel_index``.\n\n :param index: (int) The\291 \ index of the point. The coordinates are expressed as a \n single index\292 \ by using the dimensions as a mixed radix definition. For \n example,\293 \ in dimensions 42x10, the point [1, 4] is index \n 1*420 + 4*10 = 460.\n\294 \n :param dimensions (list of ints) The coordinate system.\n\n :returns: (list)\295 \ of coordinates of length ``len(dimensions)``."296 sentences:297 - "def coordinatesFromIndex(index, dimensions):\n \"\"\"\n Translate an index\298 \ into coordinates, using the given coordinate system.\n\n Similar to ``numpy.unravel_index``.\n\299 \n :param index: (int) The index of the point. The coordinates are expressed\300 \ as a \n single index by using the dimensions as a mixed radix definition.\301 \ For \n example, in dimensions 42x10, the point [1, 4] is index \n \302 \ 1*420 + 4*10 = 460.\n\n :param dimensions (list of ints) The coordinate\303 \ system.\n\n :returns: (list) of coordinates of length ``len(dimensions)``.\n\304 \ \"\"\"\n coordinates = [0] * len(dimensions)\n\n shifted = index\n for i\305 \ in xrange(len(dimensions) - 1, 0, -1):\n coordinates[i] = shifted % dimensions[i]\n\306 \ shifted = shifted / dimensions[i]\n\n coordinates[0] = shifted\n\n return\307 \ coordinates"308 - "def step(self, observation, **extra_feed):\n \"\"\"\n Compute next\309 \ action(s) given the observation(s)\n\n Parameters:\n ----------\n\310 \n observation observation data (either single or a batch)\n\n \311 \ **extra_feed additional data such as state or mask (names of the arguments\312 \ should match the ones in constructor, see __init__)\n\n Returns:\n \313 \ -------\n (action, value estimate, next state, negative log likelihood\314 \ of the action under current policy parameters) tuple\n \"\"\"\n\n \315 \ a, v, state, neglogp = self._evaluate([self.action, self.vf, self.state,\316 \ self.neglogp], observation, **extra_feed)\n if state.size == 0:\n \317 \ state = None\n return a, v, state, neglogp"318 - "def pretty_eta(seconds_left):\n \"\"\"Print the number of seconds in human\319 \ readable format.\n\n Examples:\n 2 days\n 2 hours and 37 minutes\n\320 \ less than a minute\n\n Paramters\n ---------\n seconds_left: int\n\321 \ Number of seconds to be converted to the ETA\n Returns\n -------\n\322 \ eta: str\n String representing the pretty ETA.\n \"\"\"\n minutes_left\323 \ = seconds_left // 60\n seconds_left %= 60\n hours_left = minutes_left\324 \ // 60\n minutes_left %= 60\n days_left = hours_left // 24\n hours_left\325 \ %= 24\n\n def helper(cnt, name):\n return \"{} {}{}\".format(str(cnt),\326 \ name, ('s' if cnt > 1 else ''))\n\n if days_left > 0:\n msg = helper(days_left,\327 \ 'day')\n if hours_left > 0:\n msg += ' and ' + helper(hours_left,\328 \ 'hour')\n return msg\n if hours_left > 0:\n msg = helper(hours_left,\329 \ 'hour')\n if minutes_left > 0:\n msg += ' and ' + helper(minutes_left,\330 \ 'minute')\n return msg\n if minutes_left > 0:\n return helper(minutes_left,\331 \ 'minute')\n return 'less than a minute'"332- source_sentence: Validates control dictionary for the experiment context333 sentences:334 - "def load_file_list(path=None, regx='\\.jpg', printable=True, keep_prefix=False):\n\335 \ r\"\"\"Return a file list in a folder by given a path and regular expression.\n\336 \n Parameters\n ----------\n path : str or None\n A folder path,\337 \ if `None`, use the current directory.\n regx : str\n The regx of file\338 \ name.\n printable : boolean\n Whether to print the files infomation.\n\339 \ keep_prefix : boolean\n Whether to keep path in the file name.\n\n\340 \ Examples\n ----------\n >>> file_list = tl.files.load_file_list(path=None,\341 \ regx='w1pre_[0-9]+\\.(npz)')\n\n \"\"\"\n if path is None:\n path\342 \ = os.getcwd()\n file_list = os.listdir(path)\n return_list = []\n for\343 \ _, f in enumerate(file_list):\n if re.search(regx, f):\n return_list.append(f)\n\344 \ # return_list.sort()\n if keep_prefix:\n for i, f in enumerate(return_list):\n\345 \ return_list[i] = os.path.join(path, f)\n\n if printable:\n \346 \ logging.info('Match file list = %s' % return_list)\n logging.info('Number\347 \ of files = %d' % len(return_list))\n return return_list"348 - "def getCompletingSwarms(self):\n \"\"\"Return the list of all completing swarms.\n\349 \n Parameters:\n ---------------------------------------------------------------------\n\350 \ retval: list of active swarm Ids\n \"\"\"\n swarmIds = []\n for\351 \ swarmId, info in self._state['swarms'].iteritems():\n if info['status']\352 \ == 'completing':\n swarmIds.append(swarmId)\n\n return swarmIds"353 - "def __validateExperimentControl(self, control):\n \"\"\" Validates control\354 \ dictionary for the experiment context\"\"\"\n # Validate task list\n taskList\355 \ = control.get('tasks', None)\n if taskList is not None:\n taskLabelsList\356 \ = []\n\n for task in taskList:\n validateOpfJsonValue(task, \"opfTaskSchema.json\"\357 )\n validateOpfJsonValue(task['taskControl'], \"opfTaskControlSchema.json\"\358 )\n\n taskLabel = task['taskLabel']\n\n assert isinstance(taskLabel,\359 \ types.StringTypes), \\\n \"taskLabel type: %r\" % type(taskLabel)\n\360 \ assert len(taskLabel) > 0, \"empty string taskLabel not is allowed\"\n\361 \n taskLabelsList.append(taskLabel.lower())\n\n taskLabelDuplicates\362 \ = filter(lambda x: taskLabelsList.count(x) > 1,\n \363 \ taskLabelsList)\n assert len(taskLabelDuplicates) == 0, \\\n \364 \ \"Duplcate task labels are not allowed: %s\" % taskLabelDuplicates\n\365 \n return"366pipeline_tag: sentence-similarity367library_name: sentence-transformers368---369 370# SentenceTransformer based on answerdotai/ModernBERT-base371 372This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) on the code_search_net dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.373 374## Model Details375 376### Model Description377- **Model Type:** Sentence Transformer378- **Base model:** [answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base) <!-- at revision 8949b909ec900327062f0ebf497f51aef5e6f0c8 -->379- **Maximum Sequence Length:** 4096 tokens380- **Output Dimensionality:** 768 dimensions381- **Similarity Function:** Cosine Similarity382- **Training Dataset:**383 - code_search_net384<!-- - **Language:** Unknown -->385<!-- - **License:** Unknown -->386 387### Model Sources388 389- **Documentation:** [Sentence Transformers Documentation](https://sbert.net)390- **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)391- **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)392 393### Full Model Architecture394 395```396SentenceTransformer(397 (0): Transformer({'max_seq_length': 4096, 'do_lower_case': False}) with Transformer model: ModernBertModel 398 (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})399)400```401 402## Usage403 404### Direct Usage (Sentence Transformers)405 406First install the Sentence Transformers library:407 408```bash409pip install -U sentence-transformers410```411 412Then you can load this model and run inference.413```python414from sentence_transformers import SentenceTransformer415 416# Download from the 🤗 Hub417model = SentenceTransformer("juanwisz/modernbert-python-code-retrieval")418# Run inference419sentences = [420 'Validates control dictionary for the experiment context',421 'def __validateExperimentControl(self, control):\n """ Validates control dictionary for the experiment context"""\n # Validate task list\n taskList = control.get(\'tasks\', None)\n if taskList is not None:\n taskLabelsList = []\n\n for task in taskList:\n validateOpfJsonValue(task, "opfTaskSchema.json")\n validateOpfJsonValue(task[\'taskControl\'], "opfTaskControlSchema.json")\n\n taskLabel = task[\'taskLabel\']\n\n assert isinstance(taskLabel, types.StringTypes), \\\n "taskLabel type: %r" % type(taskLabel)\n assert len(taskLabel) > 0, "empty string taskLabel not is allowed"\n\n taskLabelsList.append(taskLabel.lower())\n\n taskLabelDuplicates = filter(lambda x: taskLabelsList.count(x) > 1,\n taskLabelsList)\n assert len(taskLabelDuplicates) == 0, \\\n "Duplcate task labels are not allowed: %s" % taskLabelDuplicates\n\n return',422 'def load_file_list(path=None, regx=\'\\.jpg\', printable=True, keep_prefix=False):\n r"""Return a file list in a folder by given a path and regular expression.\n\n Parameters\n ----------\n path : str or None\n A folder path, if `None`, use the current directory.\n regx : str\n The regx of file name.\n printable : boolean\n Whether to print the files infomation.\n keep_prefix : boolean\n Whether to keep path in the file name.\n\n Examples\n ----------\n >>> file_list = tl.files.load_file_list(path=None, regx=\'w1pre_[0-9]+\\.(npz)\')\n\n """\n if path is None:\n path = os.getcwd()\n file_list = os.listdir(path)\n return_list = []\n for _, f in enumerate(file_list):\n if re.search(regx, f):\n return_list.append(f)\n # return_list.sort()\n if keep_prefix:\n for i, f in enumerate(return_list):\n return_list[i] = os.path.join(path, f)\n\n if printable:\n logging.info(\'Match file list = %s\' % return_list)\n logging.info(\'Number of files = %d\' % len(return_list))\n return return_list',423]424embeddings = model.encode(sentences)425print(embeddings.shape)426# [3, 768]427 428# Get the similarity scores for the embeddings429similarities = model.similarity(embeddings, embeddings)430print(similarities.shape)431# [3, 3]432```433 434<!--435### Direct Usage (Transformers)436 437<details><summary>Click to see the direct usage in Transformers</summary>438 439</details>440-->441 442<!--443### Downstream Usage (Sentence Transformers)444 445You can finetune this model on your own dataset.446 447<details><summary>Click to expand</summary>448 449</details>450-->451 452<!--453### Out-of-Scope Use454 455*List how the model may foreseeably be misused and address what users ought not to do with the model.*456-->457 458<!--459## Bias, Risks and Limitations460 461*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*462-->463 464<!--465### Recommendations466 467*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*468-->469 470## Training Details471 472### Training Dataset473 474#### code_search_net475 476* Dataset: code_search_net477* Size: 412,178 training samples478* Columns: <code>query</code> and <code>positive</code>479* Approximate statistics based on the first 1000 samples:480 | | query | positive |481 |:--------|:------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------|482 | type | string | string |483 | details | <ul><li>min: 4 tokens</li><li>mean: 73.72 tokens</li><li>max: 2258 tokens</li></ul> | <ul><li>min: 46 tokens</li><li>mean: 300.87 tokens</li><li>max: 3119 tokens</li></ul> |484* Samples:485 | query | positive |486 |:------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|487 | <code>Extracts the list of arguments that start with any of the specified prefix values</code> | <code>def findArgs(args, prefixes):<br> """<br> Extracts the list of arguments that start with any of the specified prefix values<br> """<br> return list([<br> arg for arg in args<br> if len([p for p in prefixes if arg.lower().startswith(p.lower())]) > 0<br> ])</code> |488 | <code>Removes any arguments in the supplied list that are contained in the specified blacklist</code> | <code>def stripArgs(args, blacklist):<br> """<br> Removes any arguments in the supplied list that are contained in the specified blacklist<br> """<br> blacklist = [b.lower() for b in blacklist]<br> return list([arg for arg in args if arg.lower() not in blacklist])</code> |489 | <code>Executes a child process and captures its output</code> | <code>def capture(command, input=None, cwd=None, shell=False, raiseOnError=False):<br> """<br> Executes a child process and captures its output<br> """<br> <br> # Attempt to execute the child process<br> proc = subprocess.Popen(command, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE, cwd=cwd, shell=shell, universal_newlines=True)<br> (stdout, stderr) = proc.communicate(input)<br> <br> # If the child process failed and we were asked to raise an exception, do so<br> if raiseOnError == True and proc.returncode != 0:<br> raise Exception(<br> 'child process ' + str(command) +<br> ' failed with exit code ' + str(proc.returncode) +<br> '\nstdout: "' + stdout + '"' +<br> '\nstderr: "' + stderr + '"'<br> )<br> <br> return CommandOutput(proc.returncode, stdout, stderr)</code> |490* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:491 ```json492 {493 "scale": 20.0,494 "similarity_fct": "cos_sim"495 }496 ```497 498### Evaluation Dataset499 500#### code_search_net501 502* Dataset: code_search_net503* Size: 23,107 evaluation samples504* Columns: <code>query</code> and <code>positive</code>505* Approximate statistics based on the first 1000 samples:506 | | query | positive |507 |:--------|:-------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|508 | type | string | string |509 | details | <ul><li>min: 5 tokens</li><li>mean: 168.27 tokens</li><li>max: 2118 tokens</li></ul> | <ul><li>min: 48 tokens</li><li>mean: 467.9 tokens</li><li>max: 4096 tokens</li></ul> |510* Samples:511 | query | positive |512 |:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|513 | <code>Train a deepq model.<br><br> Parameters<br> -------<br> env: gym.Env<br> environment to train on<br> network: string or a function<br> neural network to use as a q function approximator. If string, has to be one of the names of registered models in baselines.common.models<br> (mlp, cnn, conv_only). If a function, should take an observation tensor and return a latent variable tensor, which<br> will be mapped to the Q function heads (see build_q_func in baselines.deepq.models for details on that)<br> seed: int or None<br> prng seed. The runs with the same seed "should" give the same results. If None, no seeding is used.<br> lr: float<br> learning rate for adam optimizer<br> total_timesteps: int<br> number of env steps to optimizer for<br> buffer_size: int<br> size of the replay buffer<br> exploration_fraction: float<br> fraction of entire training period over which the exploration rate is annealed<br> exploration_final_eps: float<br> final value of ra...</code> | <code>def learn(env,<br> network,<br> seed=None,<br> lr=5e-4,<br> total_timesteps=100000,<br> buffer_size=50000,<br> exploration_fraction=0.1,<br> exploration_final_eps=0.02,<br> train_freq=1,<br> batch_size=32,<br> print_freq=100,<br> checkpoint_freq=10000,<br> checkpoint_path=None,<br> learning_starts=1000,<br> gamma=1.0,<br> target_network_update_freq=500,<br> prioritized_replay=False,<br> prioritized_replay_alpha=0.6,<br> prioritized_replay_beta0=0.4,<br> prioritized_replay_beta_iters=None,<br> prioritized_replay_eps=1e-6,<br> param_noise=False,<br> callback=None,<br> load_path=None,<br> **network_kwargs<br> ):<br> """Train a deepq model.<br><br> Parameters<br> -------<br> env: gym.Env<br> environment to train on<br> network: string or a function<br> neural network to use as a q function approximator. If string, has to be one of the ...</code> |514 | <code>Save model to a pickle located at `path`</code> | <code>def save_act(self, path=None):<br> """Save model to a pickle located at `path`"""<br> if path is None:<br> path = os.path.join(logger.get_dir(), "model.pkl")<br><br> with tempfile.TemporaryDirectory() as td:<br> save_variables(os.path.join(td, "model"))<br> arc_name = os.path.join(td, "packed.zip")<br> with zipfile.ZipFile(arc_name, 'w') as zipf:<br> for root, dirs, files in os.walk(td):<br> for fname in files:<br> file_path = os.path.join(root, fname)<br> if file_path != arc_name:<br> zipf.write(file_path, os.path.relpath(file_path, td))<br> with open(arc_name, "rb") as f:<br> model_data = f.read()<br> with open(path, "wb") as f:<br> cloudpickle.dump((model_data, self._act_params), f)</code> |515 | <code>CNN from Nature paper.</code> | <code>def nature_cnn(unscaled_images, **conv_kwargs):<br> """<br> CNN from Nature paper.<br> """<br> scaled_images = tf.cast(unscaled_images, tf.float32) / 255.<br> activ = tf.nn.relu<br> h = activ(conv(scaled_images, 'c1', nf=32, rf=8, stride=4, init_scale=np.sqrt(2),<br> **conv_kwargs))<br> h2 = activ(conv(h, 'c2', nf=64, rf=4, stride=2, init_scale=np.sqrt(2), **conv_kwargs))<br> h3 = activ(conv(h2, 'c3', nf=64, rf=3, stride=1, init_scale=np.sqrt(2), **conv_kwargs))<br> h3 = conv_to_fc(h3)<br> return activ(fc(h3, 'fc1', nh=512, init_scale=np.sqrt(2)))</code> |516* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:517 ```json518 {519 "scale": 20.0,520 "similarity_fct": "cos_sim"521 }522 ```523 524### Training Hyperparameters525#### Non-Default Hyperparameters526 527- `eval_strategy`: epoch528- `per_device_train_batch_size`: 4529- `gradient_accumulation_steps`: 4530- `learning_rate`: 2e-05531- `num_train_epochs`: 10532- `warmup_steps`: 1000533- `fp16`: True534 535#### All Hyperparameters536<details><summary>Click to expand</summary>537 538- `overwrite_output_dir`: False539- `do_predict`: False540- `eval_strategy`: epoch541- `prediction_loss_only`: True542- `per_device_train_batch_size`: 4543- `per_device_eval_batch_size`: 8544- `per_gpu_train_batch_size`: None545- `per_gpu_eval_batch_size`: None546- `gradient_accumulation_steps`: 4547- `eval_accumulation_steps`: None548- `torch_empty_cache_steps`: None549- `learning_rate`: 2e-05550- `weight_decay`: 0.0551- `adam_beta1`: 0.9552- `adam_beta2`: 0.999553- `adam_epsilon`: 1e-08554- `max_grad_norm`: 1.0555- `num_train_epochs`: 10556- `max_steps`: -1557- `lr_scheduler_type`: linear558- `lr_scheduler_kwargs`: {}559- `warmup_ratio`: 0.0560- `warmup_steps`: 1000561- `log_level`: passive562- `log_level_replica`: warning563- `log_on_each_node`: True564- `logging_nan_inf_filter`: True565- `save_safetensors`: True566- `save_on_each_node`: False567- `save_only_model`: False568- `restore_callback_states_from_checkpoint`: False569- `no_cuda`: False570- `use_cpu`: False571- `use_mps_device`: False572- `seed`: 42573- `data_seed`: None574- `jit_mode_eval`: False575- `use_ipex`: False576- `bf16`: False577- `fp16`: True578- `fp16_opt_level`: O1579- `half_precision_backend`: auto580- `bf16_full_eval`: False581- `fp16_full_eval`: False582- `tf32`: None583- `local_rank`: 0584- `ddp_backend`: None585- `tpu_num_cores`: None586- `tpu_metrics_debug`: False587- `debug`: []588- `dataloader_drop_last`: False589- `dataloader_num_workers`: 0590- `dataloader_prefetch_factor`: None591- `past_index`: -1592- `disable_tqdm`: False593- `remove_unused_columns`: True594- `label_names`: None595- `load_best_model_at_end`: False596- `ignore_data_skip`: False597- `fsdp`: []598- `fsdp_min_num_params`: 0599- `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}600- `fsdp_transformer_layer_cls_to_wrap`: None601- `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}602- `deepspeed`: None603- `label_smoothing_factor`: 0.0604- `optim`: adamw_torch605- `optim_args`: None606- `adafactor`: False607- `group_by_length`: False608- `length_column_name`: length609- `ddp_find_unused_parameters`: None610- `ddp_bucket_cap_mb`: None611- `ddp_broadcast_buffers`: False612- `dataloader_pin_memory`: True613- `dataloader_persistent_workers`: False614- `skip_memory_metrics`: True615- `use_legacy_prediction_loop`: False616- `push_to_hub`: False617- `resume_from_checkpoint`: None618- `hub_model_id`: None619- `hub_strategy`: every_save620- `hub_private_repo`: None621- `hub_always_push`: False622- `gradient_checkpointing`: False623- `gradient_checkpointing_kwargs`: None624- `include_inputs_for_metrics`: False625- `include_for_metrics`: []626- `eval_do_concat_batches`: True627- `fp16_backend`: auto628- `push_to_hub_model_id`: None629- `push_to_hub_organization`: None630- `mp_parameters`: 631- `auto_find_batch_size`: False632- `full_determinism`: False633- `torchdynamo`: None634- `ray_scope`: last635- `ddp_timeout`: 1800636- `torch_compile`: False637- `torch_compile_backend`: None638- `torch_compile_mode`: None639- `dispatch_batches`: None640- `split_batches`: None641- `include_tokens_per_second`: False642- `include_num_input_tokens_seen`: False643- `neftune_noise_alpha`: None644- `optim_target_modules`: None645- `batch_eval_metrics`: False646- `eval_on_start`: False647- `use_liger_kernel`: False648- `eval_use_gather_object`: False649- `average_tokens_across_devices`: False650- `prompts`: None651- `batch_sampler`: batch_sampler652- `multi_dataset_batch_sampler`: proportional653 654</details>655 656### Training Logs657<details><summary>Click to expand</summary>658 659| Epoch | Step | Training Loss | Validation Loss |660|:------:|:-----:|:-------------:|:---------------:|661| 0.0078 | 200 | 0.634 | - |662| 0.0155 | 400 | 0.0046 | - |663| 0.0233 | 600 | 0.0009 | - |664| 0.0311 | 800 | 0.0004 | - |665| 0.0388 | 1000 | 0.0001 | - |666| 0.0466 | 1200 | 0.0002 | - |667| 0.0543 | 1400 | 0.0001 | - |668| 0.0621 | 1600 | 0.0001 | - |669| 0.0699 | 1800 | 0.0001 | - |670| 0.0776 | 2000 | 0.0 | - |671| 0.0854 | 2200 | 0.0 | - |672| 0.0932 | 2400 | 0.0 | - |673| 0.1009 | 2600 | 0.0 | - |674| 0.1087 | 2800 | 0.0005 | - |675| 0.1165 | 3000 | 0.0005 | - |676| 0.1242 | 3200 | 0.0002 | - |677| 0.1320 | 3400 | 0.0 | - |678| 0.1397 | 3600 | 0.0 | - |679| 0.1475 | 3800 | 0.0 | - |680| 0.1553 | 4000 | 0.0001 | - |681| 0.1630 | 4200 | 0.0 | - |682| 0.1708 | 4400 | 0.0001 | - |683| 0.1786 | 4600 | 0.0001 | - |684| 0.1863 | 4800 | 0.0 | - |685| 0.1941 | 5000 | 0.0 | - |686| 0.2019 | 5200 | 0.0 | - |687| 0.2096 | 5400 | 0.0 | - |688| 0.2174 | 5600 | 0.0 | - |689| 0.2251 | 5800 | 0.0 | - |690| 0.2329 | 6000 | 0.0004 | - |691| 0.2407 | 6200 | 0.0 | - |692| 0.2484 | 6400 | 0.0001 | - |693| 0.2562 | 6600 | 0.0 | - |694| 0.2640 | 6800 | 0.0 | - |695| 0.2717 | 7000 | 0.0 | - |696| 0.2795 | 7200 | 0.0 | - |697| 0.2873 | 7400 | 0.0 | - |698| 0.2950 | 7600 | 0.0 | - |699| 0.3028 | 7800 | 0.0 | - |700| 0.3105 | 8000 | 0.0 | - |701| 0.3183 | 8200 | 0.0 | - |702| 0.3261 | 8400 | 0.0004 | - |703| 0.3338 | 8600 | 0.0 | - |704| 0.3416 | 8800 | 0.0 | - |705| 0.3494 | 9000 | 0.0 | - |706| 0.3571 | 9200 | 0.0 | - |707| 0.3649 | 9400 | 0.0 | - |708| 0.3727 | 9600 | 0.0 | - |709| 0.3804 | 9800 | 0.0 | - |710| 0.3882 | 10000 | 0.0 | - |711| 0.3959 | 10200 | 0.0 | - |712| 0.4037 | 10400 | 0.0 | - |713| 0.4115 | 10600 | 0.0 | - |714| 0.4192 | 10800 | 0.0 | - |715| 0.4270 | 11000 | 0.0 | - |716| 0.4348 | 11200 | 0.0 | - |717| 0.4425 | 11400 | 0.0 | - |718| 0.4503 | 11600 | 0.0 | - |719| 0.4581 | 11800 | 0.0 | - |720| 0.4658 | 12000 | 0.0 | - |721| 0.4736 | 12200 | 0.0 | - |722| 0.4813 | 12400 | 0.0 | - |723| 0.4891 | 12600 | 0.0005 | - |724| 0.4969 | 12800 | 0.0 | - |725| 0.5046 | 13000 | 0.0 | - |726| 0.5124 | 13200 | 0.0001 | - |727| 0.5202 | 13400 | 0.0 | - |728| 0.5279 | 13600 | 0.0 | - |729| 0.5357 | 13800 | 0.0 | - |730| 0.5435 | 14000 | 0.0 | - |731| 0.5512 | 14200 | 0.0 | - |732| 0.5590 | 14400 | 0.0004 | - |733| 0.5667 | 14600 | 0.0 | - |734| 0.5745 | 14800 | 0.0 | - |735| 0.5823 | 15000 | 0.0 | - |736| 0.5900 | 15200 | 0.0 | - |737| 0.5978 | 15400 | 0.0 | - |738| 0.6056 | 15600 | 0.0 | - |739| 0.6133 | 15800 | 0.0 | - |740| 0.6211 | 16000 | 0.0 | - |741| 0.6289 | 16200 | 0.0 | - |742| 0.6366 | 16400 | 0.0006 | - |743| 0.6444 | 16600 | 0.0 | - |744| 0.6521 | 16800 | 0.0005 | - |745| 0.6599 | 17000 | 0.0 | - |746| 0.6677 | 17200 | 0.0 | - |747| 0.6754 | 17400 | 0.0 | - |748| 0.6832 | 17600 | 0.0 | - |749| 0.6910 | 17800 | 0.0 | - |750| 0.6987 | 18000 | 0.0005 | - |751| 0.7065 | 18200 | 0.0001 | - |752| 0.7143 | 18400 | 0.0 | - |753| 0.7220 | 18600 | 0.0 | - |754| 0.7298 | 18800 | 0.0 | - |755| 0.7375 | 19000 | 0.0 | - |756| 0.7453 | 19200 | 0.0 | - |757| 0.7531 | 19400 | 0.0 | - |758| 0.7608 | 19600 | 0.0 | - |759| 0.7686 | 19800 | 0.0001 | - |760| 0.7764 | 20000 | 0.0 | - |761| 0.7841 | 20200 | 0.0 | - |762| 0.7919 | 20400 | 0.0 | - |763| 0.7997 | 20600 | 0.0004 | - |764| 0.8074 | 20800 | 0.0 | - |765| 0.8152 | 21000 | 0.0 | - |766| 0.8229 | 21200 | 0.0 | - |767| 0.8307 | 21400 | 0.0009 | - |768| 0.8385 | 21600 | 0.0 | - |769| 0.8462 | 21800 | 0.0 | - |770| 0.8540 | 22000 | 0.0 | - |771| 0.8618 | 22200 | 0.0 | - |772| 0.8695 | 22400 | 0.0002 | - |773| 0.8773 | 22600 | 0.0 | - |774| 0.8851 | 22800 | 0.0 | - |775| 0.8928 | 23000 | 0.0001 | - |776| 0.9006 | 23200 | 0.0 | - |777| 0.9083 | 23400 | 0.0 | - |778| 0.9161 | 23600 | 0.0 | - |779| 0.9239 | 23800 | 0.0 | - |780| 0.9316 | 24000 | 0.0 | - |781| 0.9394 | 24200 | 0.0 | - |782| 0.9472 | 24400 | 0.0 | - |783| 0.9549 | 24600 | 0.0 | - |784| 0.9627 | 24800 | 0.0 | - |785| 0.9704 | 25000 | 0.0 | - |786| 0.9782 | 25200 | 0.0 | - |787| 0.9860 | 25400 | 0.0 | - |788| 0.9937 | 25600 | 0.0 | - |789| 1.0 | 25762 | - | 0.0001 |790| 1.0015 | 25800 | 0.0005 | - |791| 1.0092 | 26000 | 0.0 | - |792| 1.0170 | 26200 | 0.0 | - |793| 1.0248 | 26400 | 0.0 | - |794| 1.0325 | 26600 | 0.0 | - |795| 1.0403 | 26800 | 0.0 | - |796| 1.0481 | 27000 | 0.0 | - |797| 1.0558 | 27200 | 0.0 | - |798| 1.0636 | 27400 | 0.0 | - |799| 1.0713 | 27600 | 0.0 | - |800| 1.0791 | 27800 | 0.0 | - |801| 1.0869 | 28000 | 0.0 | - |802| 1.0946 | 28200 | 0.0 | - |803| 1.1024 | 28400 | 0.0 | - |804| 1.1102 | 28600 | 0.0 | - |805| 1.1179 | 28800 | 0.0 | - |806| 1.1257 | 29000 | 0.0 | - |807| 1.1335 | 29200 | 0.0 | - |808| 1.1412 | 29400 | 0.0 | - |809| 1.1490 | 29600 | 0.0 | - |810| 1.1567 | 29800 | 0.0 | - |811| 1.1645 | 30000 | 0.0 | - |812| 1.1723 | 30200 | 0.0 | - |813| 1.1800 | 30400 | 0.0 | - |814| 1.1878 | 30600 | 0.0 | - |815| 1.1956 | 30800 | 0.0 | - |816| 1.2033 | 31000 | 0.0 | - |817| 1.2111 | 31200 | 0.0 | - |818| 1.2189 | 31400 | 0.0 | - |819| 1.2266 | 31600 | 0.0004 | - |820| 1.2344 | 31800 | 0.0004 | - |821| 1.2421 | 32000 | 0.0 | - |822| 1.2499 | 32200 | 0.0 | - |823| 1.2577 | 32400 | 0.0 | - |824| 1.2654 | 32600 | 0.0 | - |825| 1.2732 | 32800 | 0.0 | - |826| 1.2810 | 33000 | 0.0 | - |827| 1.2887 | 33200 | 0.0 | - |828| 1.2965 | 33400 | 0.0 | - |829| 1.3043 | 33600 | 0.0 | - |830| 1.3120 | 33800 | 0.0 | - |831| 1.3198 | 34000 | 0.0 | - |832| 1.3275 | 34200 | 0.0 | - |833| 1.3353 | 34400 | 0.0 | - |834| 1.3431 | 34600 | 0.0 | - |835| 1.3508 | 34800 | 0.0004 | - |836| 1.3586 | 35000 | 0.0005 | - |837| 1.3664 | 35200 | 0.0004 | - |838| 1.3741 | 35400 | 0.0011 | - |839| 1.3819 | 35600 | 0.0 | - |840| 1.3897 | 35800 | 0.0 | - |841| 1.3974 | 36000 | 0.0 | - |842| 1.4052 | 36200 | 0.0 | - |843| 1.4129 | 36400 | 0.0 | - |844| 1.4207 | 36600 | 0.0 | - |845| 1.4285 | 36800 | 0.0 | - |846| 1.4362 | 37000 | 0.0 | - |847| 1.4440 | 37200 | 0.0001 | - |848| 1.4518 | 37400 | 0.0 | - |849| 1.4595 | 37600 | 0.0 | - |850| 1.4673 | 37800 | 0.0 | - |851| 1.4751 | 38000 | 0.0 | - |852| 1.4828 | 38200 | 0.0004 | - |853| 1.4906 | 38400 | 0.0003 | - |854| 1.4983 | 38600 | 0.0 | - |855| 1.5061 | 38800 | 0.0 | - |856| 1.5139 | 39000 | 0.0 | - |857| 1.5216 | 39200 | 0.0 | - |858| 1.5294 | 39400 | 0.0004 | - |859| 1.5372 | 39600 | 0.0004 | - |860| 1.5449 | 39800 | 0.0 | - |861| 1.5527 | 40000 | 0.0 | - |862| 1.5605 | 40200 | 0.0 | - |863| 1.5682 | 40400 | 0.0 | - |864| 1.5760 | 40600 | 0.0009 | - |865| 1.5837 | 40800 | 0.0 | - |866| 1.5915 | 41000 | 0.0009 | - |867| 1.5993 | 41200 | 0.0 | - |868| 1.6070 | 41400 | 0.0 | - |869| 1.6148 | 41600 | 0.0 | - |870| 1.6226 | 41800 | 0.0 | - |871| 1.6303 | 42000 | 0.0 | - |872| 1.6381 | 42200 | 0.0 | - |873| 1.6459 | 42400 | 0.0 | - |874| 1.6536 | 42600 | 0.0 | - |875| 1.6614 | 42800 | 0.0 | - |876| 1.6691 | 43000 | 0.0 | - |877| 1.6769 | 43200 | 0.0 | - |878| 1.6847 | 43400 | 0.0 | - |879| 1.6924 | 43600 | 0.0 | - |880| 1.7002 | 43800 | 0.0 | - |881| 1.7080 | 44000 | 0.0 | - |882| 1.7157 | 44200 | 0.0 | - |883| 1.7235 | 44400 | 0.0 | - |884| 1.7313 | 44600 | 0.0 | - |885| 1.7390 | 44800 | 0.0 | - |886| 1.7468 | 45000 | 0.0 | - |887| 1.7545 | 45200 | 0.0 | - |888| 1.7623 | 45400 | 0.0 | - |889| 1.7701 | 45600 | 0.0 | - |890| 1.7778 | 45800 | 0.0 | - |891| 1.7856 | 46000 | 0.0 | - |892| 1.7934 | 46200 | 0.0 | - |893| 1.8011 | 46400 | 0.0 | - |894| 1.8089 | 46600 | 0.0 | - |895| 1.8167 | 46800 | 0.0 | - |896| 1.8244 | 47000 | 0.0 | - |897| 1.8322 | 47200 | 0.0 | - |898| 1.8399 | 47400 | 0.0 | - |899| 1.8477 | 47600 | 0.0 | - |900| 1.8555 | 47800 | 0.0004 | - |901| 1.8632 | 48000 | 0.0 | - |902| 1.8710 | 48200 | 0.0 | - |903| 1.8788 | 48400 | 0.0 | - |904| 1.8865 | 48600 | 0.0 | - |905| 1.8943 | 48800 | 0.0 | - |906| 1.9021 | 49000 | 0.0004 | - |907| 1.9098 | 49200 | 0.0 | - |908| 1.9176 | 49400 | 0.0 | - |909| 1.9253 | 49600 | 0.0004 | - |910| 1.9331 | 49800 | 0.0 | - |911| 1.9409 | 50000 | 0.0 | - |912| 1.9486 | 50200 | 0.0 | - |913| 1.9564 | 50400 | 0.0 | - |914| 1.9642 | 50600 | 0.0004 | - |915| 1.9719 | 50800 | 0.0 | - |916| 1.9797 | 51000 | 0.0 | - |917| 1.9875 | 51200 | 0.0 | - |918| 1.9952 | 51400 | 0.0004 | - |919| 2.0 | 51524 | - | 0.0001 |920| 2.0030 | 51600 | 0.0 | - |921| 2.0107 | 51800 | 0.0 | - |922| 2.0185 | 52000 | 0.0 | - |923| 2.0262 | 52200 | 0.0 | - |924| 2.0340 | 52400 | 0.0004 | - |925| 2.0418 | 52600 | 0.0004 | - |926| 2.0495 | 52800 | 0.0 | - |927| 2.0573 | 53000 | 0.0008 | - |928| 2.0651 | 53200 | 0.0 | - |929| 2.0728 | 53400 | 0.0 | - |930| 2.0806 | 53600 | 0.0 | - |931| 2.0883 | 53800 | 0.0 | - |932| 2.0961 | 54000 | 0.0 | - |933| 2.1039 | 54200 | 0.0 | - |934| 2.1116 | 54400 | 0.0 | - |935| 2.1194 | 54600 | 0.0 | - |936| 2.1272 | 54800 | 0.0 | - |937| 2.1349 | 55000 | 0.0 | - |938| 2.1427 | 55200 | 0.0 | - |939| 2.1505 | 55400 | 0.0 | - |940| 2.1582 | 55600 | 0.0 | - |941| 2.1660 | 55800 | 0.0 | - |942| 2.1737 | 56000 | 0.0 | - |943| 2.1815 | 56200 | 0.0 | - |944| 2.1893 | 56400 | 0.0 | - |945| 2.1970 | 56600 | 0.0 | - |946| 2.2048 | 56800 | 0.0 | - |947| 2.2126 | 57000 | 0.0 | - |948| 2.2203 | 57200 | 0.0 | - |949| 2.2281 | 57400 | 0.0 | - |950| 2.2359 | 57600 | 0.0 | - |951| 2.2436 | 57800 | 0.0 | - |952| 2.2514 | 58000 | 0.0004 | - |953| 2.2591 | 58200 | 0.0 | - |954| 2.2669 | 58400 | 0.0004 | - |955| 2.2747 | 58600 | 0.0 | - |956| 2.2824 | 58800 | 0.0 | - |957| 2.2902 | 59000 | 0.0 | - |958| 2.2980 | 59200 | 0.0 | - |959| 2.3057 | 59400 | 0.0 | - |960| 2.3135 | 59600 | 0.0 | - |961| 2.3213 | 59800 | 0.0004 | - |962| 2.3290 | 60000 | 0.0 | - |963| 2.3368 | 60200 | 0.0004 | - |964| 2.3445 | 60400 | 0.0 | - |965| 2.3523 | 60600 | 0.0 | - |966| 2.3601 | 60800 | 0.0 | - |967| 2.3678 | 61000 | 0.0 | - |968| 2.3756 | 61200 | 0.0 | - |969| 2.3834 | 61400 | 0.0 | - |970| 2.3911 | 61600 | 0.0 | - |971| 2.3989 | 61800 | 0.0 | - |972| 2.4067 | 62000 | 0.0005 | - |973| 2.4144 | 62200 | 0.0 | - |974| 2.4222 | 62400 | 0.0 | - |975| 2.4299 | 62600 | 0.0 | - |976| 2.4377 | 62800 | 0.0 | - |977| 2.4455 | 63000 | 0.0 | - |978| 2.4532 | 63200 | 0.0 | - |979| 2.4610 | 63400 | 0.0 | - |980| 2.4688 | 63600 | 0.0 | - |981| 2.4765 | 63800 | 0.0 | - |982| 2.4843 | 64000 | 0.0 | - |983| 2.4921 | 64200 | 0.0 | - |984| 2.4998 | 64400 | 0.0 | - |985| 2.5076 | 64600 | 0.0 | - |986| 2.5153 | 64800 | 0.0 | - |987| 2.5231 | 65000 | 0.0 | - |988| 2.5309 | 65200 | 0.0 | - |989| 2.5386 | 65400 | 0.0 | - |990| 2.5464 | 65600 | 0.0004 | - |991| 2.5542 | 65800 | 0.0 | - |992| 2.5619 | 66000 | 0.0 | - |993| 2.5697 | 66200 | 0.0 | - |994| 2.5775 | 66400 | 0.0 | - |995| 2.5852 | 66600 | 0.0 | - |996| 2.5930 | 66800 | 0.0 | - |997| 2.6007 | 67000 | 0.0 | - |998| 2.6085 | 67200 | 0.0 | - |999| 2.6163 | 67400 | 0.0 | - |1000| 2.6240 | 67600 | 0.0 | - |1001| 2.6318 | 67800 | 0.0 | - |1002| 2.6396 | 68000 | 0.0 | - |1003| 2.6473 | 68200 | 0.0 | - |1004| 2.6551 | 68400 | 0.0 | - |1005| 2.6629 | 68600 | 0.0 | - |1006| 2.6706 | 68800 | 0.0004 | - |1007| 2.6784 | 69000 | 0.0 | - |1008| 2.6861 | 69200 | 0.0 | - |1009| 2.6939 | 69400 | 0.0 | - |1010| 2.7017 | 69600 | 0.0004 | - |1011| 2.7094 | 69800 | 0.0004 | - |1012| 2.7172 | 70000 | 0.0 | - |1013| 2.7250 | 70200 | 0.0 | - |1014| 2.7327 | 70400 | 0.0 | - |1015| 2.7405 | 70600 | 0.0 | - |1016| 2.7483 | 70800 | 0.0 | - |1017| 2.7560 | 71000 | 0.0004 | - |1018| 2.7638 | 71200 | 0.0 | - |1019| 2.7715 | 71400 | 0.0 | - |1020| 2.7793 | 71600 | 0.0 | - |1021| 2.7871 | 71800 | 0.0 | - |1022| 2.7948 | 72000 | 0.0 | - |1023| 2.8026 | 72200 | 0.0 | - |1024| 2.8104 | 72400 | 0.0 | - |1025| 2.8181 | 72600 | 0.0 | - |1026| 2.8259 | 72800 | 0.0 | - |1027| 2.8337 | 73000 | 0.0004 | - |1028| 2.8414 | 73200 | 0.0 | - |1029| 2.8492 | 73400 | 0.0 | - |1030| 2.8569 | 73600 | 0.0 | - |1031| 2.8647 | 73800 | 0.0004 | - |1032| 2.8725 | 74000 | 0.0 | - |1033| 2.8802 | 74200 | 0.0 | - |1034| 2.8880 | 74400 | 0.0 | - |1035| 2.8958 | 74600 | 0.0 | - |1036| 2.9035 | 74800 | 0.0 | - |1037| 2.9113 | 75000 | 0.0 | - |1038| 2.9191 | 75200 | 0.0 | - |1039| 2.9268 | 75400 | 0.0004 | - |1040| 2.9346 | 75600 | 0.0 | - |1041| 2.9423 | 75800 | 0.0 | - |1042| 2.9501 | 76000 | 0.0 | - |1043| 2.9579 | 76200 | 0.0 | - |1044| 2.9656 | 76400 | 0.0 | - |1045| 2.9734 | 76600 | 0.0004 | - |1046| 2.9812 | 76800 | 0.0 | - |1047| 2.9889 | 77000 | 0.0 | - |1048| 2.9967 | 77200 | 0.0 | - |1049| 3.0 | 77286 | - | 0.0000 |1050 1051</details>1052 1053### Framework Versions1054- Python: 3.11.111055- Sentence Transformers: 3.3.11056- Transformers: 4.48.01057- PyTorch: 2.5.1+cu1211058- Accelerate: 1.2.11059- Datasets: 3.2.01060- Tokenizers: 0.21.01061 1062## Citation1063 1064### BibTeX1065 1066#### ModernBERT1067```bibtex1068@misc{warner2024smarterbetterfasterlonger,1069 title={Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inference}, 1070 author={Benjamin Warner and Antoine Chaffin and Benjamin Clavié and Orion Weller and Oskar Hallström and Said Taghadouini and Alexis Gallagher and Raja Biswas and Faisal Ladhak and Tom Aarsen and Nathan Cooper and Griffin Adams and Jeremy Howard and Iacopo Poli},1071 year={2024},1072 eprint={2412.13663},1073 archivePrefix={arXiv},1074 primaryClass={cs.CL},1075 url={https://arxiv.org/abs/2412.13663}, 1076}1077```1078 1079#### Sentence Transformers1080```bibtex1081@inproceedings{reimers-2019-sentence-bert,1082 title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",1083 author = "Reimers, Nils and Gurevych, Iryna",1084 booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",1085 month = "11",1086 year = "2019",1087 publisher = "Association for Computational Linguistics",1088 url = "https://arxiv.org/abs/1908.10084",1089}1090```1091 1092#### MultipleNegativesRankingLoss1093```bibtex1094@misc{henderson2017efficient,1095 title={Efficient Natural Language Response Suggestion for Smart Reply},1096 author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},1097 year={2017},1098 eprint={1705.00652},1099 archivePrefix={arXiv},1100 primaryClass={cs.CL}1101}1102```1103 1104<!--1105## Glossary1106 1107*Clearly define terms in order to be accessible across audiences.*1108-->1109 1110<!--1111## Model Card Authors1112 1113*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*1114-->1115 1116<!--1117## Model Card Contact1118 1119*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*1120-->